Base_JSON/prefixO/json/O11/O11-3003.json

{
    "paper_id": "O11-3003",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T08:05:27.931683Z"
    },
    "title": "Some Chances and Challenges in Applying Language Technologies to Historical Studies in Chinese",
    "authors": [
        {
            "first": "Chao-Lin",
            "middle": [],
            "last": "Liu",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "National Chengchi University",
                "location": {
                    "country": "Taiwan"
                }
            },
            "email": ""
        },
        {
            "first": "Jin",
            "middle": [
                "+"
            ],
            "last": "Guantao",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Qingfeng",
            "middle": [],
            "last": "Liu",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Wei-Yun",
            "middle": [],
            "last": "Chiu",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Yih-Soong",
            "middle": [],
            "last": "Yu",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "National Chengchi University",
                "location": {
                    "country": "Taiwan"
                }
            },
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "We report applications of language technology to analyzing historical documents in the Database for the Study of Modern Chinese Thoughts and Literature (DSMCTL). We studied two historical issues with the reported techniques: the conceptualization of \"huaren\" (\u83ef \u4eba , Chinese people) and the attempt to institute constitutional monarchy in the late Qing dynasty. We also discuss research challenges for supporting sophisticated issues using our experience with DSMCTL, the Database of Government Officials of the Republic of China, and the Dream of the Red Chamber. Advanced techniques and tools for lexical, syntactic, semantic, and pragmatic processing of language information, along with more thorough data collection, are needed to strengthen the collaboration between historians and computer scientists.",
    "pdf_parse": {
        "paper_id": "O11-3003",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "We report applications of language technology to analyzing historical documents in the Database for the Study of Modern Chinese Thoughts and Literature (DSMCTL). We studied two historical issues with the reported techniques: the conceptualization of \"huaren\" (\u83ef \u4eba , Chinese people) and the attempt to institute constitutional monarchy in the late Qing dynasty. We also discuss research challenges for supporting sophisticated issues using our experience with DSMCTL, the Database of Government Officials of the Republic of China, and the Dream of the Red Chamber. Advanced techniques and tools for lexical, syntactic, semantic, and pragmatic processing of language information, along with more thorough data collection, are needed to strengthen the collaboration between historians and computer scientists.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "Natural language processing (NLP) is a well-known research area in computer science and has been successfully applied to handle and analyze modern textual material in the past decades.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1."
            },
            {
                "text": "Applying Language Technologies to Historical Studies in Chinese techniques of information retrieval to support the study.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Some Chances and Challenges in 29",
                "sec_num": null
            },
            {
                "text": "We also studied the attitude of the Qing government towards the Chinese workers who worked in other countries between 1875 and 1911. We analyzed the co-occurrences, i.e., collocations, of the keywords over the years of interest, using the documents recorded in the diplomatic documents of the late Qing dynasty. 6 Detailed observations and discussions of this historical research are reported in two other papers (Jin et al., 2011; Jin et al., 2012) that will be presented in the Third Conference of Digital Archives and Digital Humanities.",
                "cite_spans": [
                    {
                        "start": 312,
                        "end": 313,
                        "text": "6",
                        "ref_id": null
                    },
                    {
                        "start": 413,
                        "end": 431,
                        "text": "(Jin et al., 2011;",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 432,
                        "end": 449,
                        "text": "Jin et al., 2012)",
                        "ref_id": "BIBREF10"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Some Chances and Challenges in 29",
                "sec_num": null
            },
            {
                "text": "While we have applied NLP techniques to support historical studies, we have also experienced some challenging problems at the lexical, syntactic, semantic, and pragmatic levels. For instance, what are the most appropriate computational functions that support a certain research need? Are the current databases good enough? We elaborate on these challenges based on our experience with the three data sources, i.e., DSMCTL, DGOROC, and DRC. No one may expect that NLP techniques will replace the major role of historians in historical studies, but the techniques should be able to work with historians to make their studies more efficient and more effective. Empirical experience reported in this paper and the literature have demonstrated the potential of NLP techniques. With the help of computing technology, historians can delegate some search work and basic analysis to computers and spend more time on higher-level philosophical issues than before.",
                "cite_spans": [
                    {
                        "start": 415,
                        "end": 439,
                        "text": "DSMCTL, DGOROC, and DRC.",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Some Chances and Challenges in 29",
                "sec_num": null
            },
            {
                "text": "The Database for the Study of Modern Chinese Thoughts and Literature contains six genres of text material that were published between 1830 and 1930. Except for the first category, most of them were collected from the late Qing dynasty: modern periodicals, personal publications of the literati, diplomatic documents, newspapers, official documents, and translated works by western commissioners. Currently, the database contains more than 120 million simplified Chinese characters. 7",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "6 \u6e05\u5b63\u5916\u4ea4\u53f2\uf9be (qing1 ji4 wai4 jiao1 shi3 liao4): http://zh.wikisource.org/zh-hant/\u6e05\u5b63\u5916\u4ea4\u53f2\uf9be\u9078\u8f2f 7 DSMCTL was first built in a project led by Guantao Jin and Qingfeng Liu while they were with the Chinese University of Hong Kong. Due to budget constraints, the historical documents were sent to China, where the simplified Chinese was used, to be scanned and entered into computers. Hence, the earliest version of DSMCTL was in simplified Chinese. A traditional Chinese version of DSMCTL is still under development.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "For modern Chinese information processing with NLP techniques, researchers rely on good machine readable lexicons and good methods to segment Chinese strings into Chinese words. Both of these infrastructural facilities are missing for the processing of non-modern Chinese text. Hence, we bootstrapped our work by computing frequent Chinese strings with the PAT Tree technique in the documents, and we asked historians to select relevant words from the frequent strings. Table 1 shows the statistics about five collections in the DSMCTL database: Constitution (\u6e05\u672b\uf9f7\u61b2\u6a94\u6848), Diplomacy (\u6e05\u5b63\u5916\u4ea4\u53f2\uf9be), Min_Bow (\u6c11\u5831), 8 Nations (\u6d77\u570b\u5716\u5fd7), 9 and New_People (\u65b0\u6c11\u53e2\u5831). 10 They contain about 11 million characters, about one tenth of the whole DSMCTL database. We refer to strings that occurred more than 10 times 11 in a collection as pseudowords. Many of these pseudowords have specific meanings, but not all of them do.",
                "cite_spans": [
                    {
                        "start": 621,
                        "end": 622,
                        "text": "9",
                        "ref_id": null
                    }
                ],
                "ref_spans": [
                    {
                        "start": 470,
                        "end": 477,
                        "text": "Table 1",
                        "ref_id": "TABREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "We ranked the pseudowords based on their frequencies, i.e., the most and the second most frequent pseudowords were ranked first and second, respectively. Then, we computed the logarithmic values of the ranks and frequencies, resulting in the curves in Figure 1 . The curves in Figure 1 indicate that the pseudowords in the Chinese historical documents, like documents written in modern English and Chinese languages (Ha et al., 2003; Xiao, 2008) , conform to Zipf's law quite well (Zipf, 1949 ).",
                "cite_spans": [
                    {
                        "start": 416,
                        "end": 433,
                        "text": "(Ha et al., 2003;",
                        "ref_id": "BIBREF5"
                    },
                    {
                        "start": 434,
                        "end": 445,
                        "text": "Xiao, 2008)",
                        "ref_id": "BIBREF20"
                    },
                    {
                        "start": 481,
                        "end": 492,
                        "text": "(Zipf, 1949",
                        "ref_id": "BIBREF21"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 252,
                        "end": 260,
                        "text": "Figure 1",
                        "ref_id": "FIGREF0"
                    },
                    {
                        "start": 277,
                        "end": 285,
                        "text": "Figure 1",
                        "ref_id": "FIGREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "8 \u6c11\u5831 (min2 bao4): http://zh.wikipedia.org/wiki/\u6c11\u5831 9 \u6d77\u570b\u5716\u5fd7 (hai3 guo2 tu2 zhi4): http://zh.wikipedia.org/wiki/\u6d77\u570b\u5716\u5fd7 10 \u65b0\u6c11\u53e2\u5831 (sin1 min2 cong2 bao4): http://zh.wikipedia.org/wiki/\u65b0\u6c11\u53e2\u5831 11 The selection of 10 as the threshold was by the historians. The choice was heuristic but arbitrary. Let and denote the rank and frequency of a word in a collection of text, respectively, Zipf's law predicts that the product of and is a constant, , as shown in Equation (1).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "(1)",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "Hence, we will observe curves that are almost straight lines after we take the logarithm (usually abbreviated as \"log\") on both sides of Equation (1) to become Equation (2). In Figure  1 , the log values of pseudoword frequencies are on the vertical axis, and the log values of the pseudoword ranks are on the horizontal axis.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 177,
                        "end": 186,
                        "text": "Figure  1",
                        "ref_id": "FIGREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "log log log",
                        "eq_num": "(2)"
                    }
                ],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "Let denote the total number of characters in a collection. We divided the word frequencies by the sizes of individual collections. In Figure 2 , the vertical axis shows the log values of the pseudoword frequencies divided by , namely, log . The curves for the distributions of the pseudowords almost overlap, suggesting that Zipf's law applied to the five collections quite uniformly, after we considered the influences of the sizes of collections.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 134,
                        "end": 142,
                        "text": "Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "The decision to divide term frequency, f, by the corpus size, N, was arbitrary, but it was very interesting to find that curves in Figure 2 almost overlap as a result. Evidently, sizes of corpora affected the shapes and positions of the Zipfian curves. Xiao (2009) attempted to study the influences of corpus size over the Zipfian curves. In one of the reported studies, Xiao sampled five small datasets of almost the same size from the General Contemporary Chinese Corpus, which contained approximately one billion Chinese characters. Zipfian curves drawn for these datasets overlapped almost perfectly.",
                "cite_spans": [
                    {
                        "start": 253,
                        "end": 264,
                        "text": "Xiao (2009)",
                        "ref_id": "BIBREF20"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 131,
                        "end": 139,
                        "text": "Figure 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Zipf's Law Applicability",
                "sec_num": "2."
            },
            {
                "text": "We examined the pseudowords and selected those that are potentially relevant to historical issues as keywords. We computed the annual and total frequencies of each of these keywords and computed the total number of keywords in each year.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "The \"Total\" curve serves as the basis for the analysis of importance of keywords. Let t 1905 , t 1906 , t 1907 , t 1908 , t 1909 , t 1910 , and t 1911 denote the total number of keywords appearing in 1905, 1906, 1907, 1908, 1909, 1910, and 1911, respectively . We could compute the total number of keywords in Constitution, T, using the following equation.",
                "cite_spans": [
                    {
                        "start": 200,
                        "end": 258,
                        "text": "1905, 1906, 1907, 1908, 1909, 1910, and 1911, respectively",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "( 3)Using the years on the horizontal axis and the annual percentage, , on the vertical axis, we analyzed the keywords in Constitution (cf. Table 1) to obtain the \"Total\" curve in Figure 3 .",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 180,
                        "end": 188,
                        "text": "Figure 3",
                        "ref_id": "FIGREF1"
                    }
                ],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "Analogously, let K denote the total number of a particular keyword, e.g., \"\u5b98\u5236,\" (guan1 zhi4, bureaucracy ) that appeared in Constitution and k n denote the number of instances the keyword appeared in a year n. We can draw a curve of annual percentage for a keyword. When the keywords appeared more frequently in a year, a historical event typically coincided with the increase in frequency (Jin et al., 2011) . We considered the keyword to be special in the year n, if",
                "cite_spans": [
                    {
                        "start": 390,
                        "end": 408,
                        "text": "(Jin et al., 2011)",
                        "ref_id": "BIBREF11"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "k n K \u03bb t n T",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": ". In Jin et al. (2011), we chose \u03bb 1.1 arbitrarily, but the selection of \u03bb can be adjusted as needed in a computer-assisted document analysis environment.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "For instance, in 1906, for \"\u5b98\u5236\" was about 0.45, and was less than 0.25. In 1907, for \"\uf9f7\u5baa\" 12 was about 0.40, and was less than 0.33. Both \"\u5b98\u5236\" and \"\uf9f7 \u5baa \" qualified as special. In 1906 and 1907, the Qing government began to consider constitutional monarchy seriously, so government officials intensively discussed the issues of \"bureaucracy\" (\"\u5b98\u5236\") and \"constitutionality\" (\"\uf9f7\u5baa\") for running the new form of government. Hence, keywords like \"\u5b98\u5236\" and \"\uf9f7\u5baa\" appeared in the emperor's memorials more often than in other years.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "In 1908, for \"\u9009\u4e3e\" 13 was about 0.45, and was less than 0.15. In fact, in 1908, the keywords about election (\"\u9009\u4e3e\" and \"\u7ae0\u7a0b\") were used more frequently in the emperor's memorials.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "After years of discussion on the fundamental issues of a constitutional monarchy, the Qing government appeared to be prepared for the new form of government and was taking steps for its realization. In 1909 and 1910, words relating to self-governance (\"\u7b79\u529e\" and \"\u81ea The temporal relationship between these six keywords' emerging importance further suggested the progression toward the establishment of a constitutional monarchy before the overthrow of the emperor in late 1911. Namely, the focus of discussion shifted from planning and preparation to realization and action.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "Our approach is more appropriate for historical studies than the Google Trends 15 approach, although the difference is subtle and may appear minuscule. The analysis of occurrences of an individual keyword, like in Google Trends, is useful for studying the changing importance of a keyword over a period of time. Evaluating the chronicle change of importance of a keyword is certainly important, but we further compare the chronicle changes of multiple keywords, which allows us to visualize the trends more directly.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Chronicle Trends of Multiple Keywords",
                "sec_num": "3."
            },
            {
                "text": "A collocation is formed by two keywords that appeared \"close\" to each other in a statement. A collocation carries more specific semantic information than an individual keyword. The occurrence of the keyword \"Chinese labor\" (\"\u534e\u5de5,\" hua2 gong1) could have referred to anything about Chinese labor, e.g., limiting (\"\u9650\u5236,\" xian4 zhi4) or protecting (\"\u4fdd\u62a4,\" bao3 hu4) the Chinese labor, while a collocation \"protect the Chinese labor\" (\"\u4fdd\u62a4\" and \"\u534e\u5de5\") provides more specific meaning than the individual keywords.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Temporal Analysis of Important Collocations",
                "sec_num": "4."
            },
            {
                "text": "Nevertheless, given that there were neither word boundaries nor sentence boundaries in pre-modern Chinese documents. We chose to define \"close\" based only on the \"distance\" between two keywords.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Temporal Analysis of Important Collocations",
                "sec_num": "4."
            },
            {
                "text": "A keyword was considered to be collocated with another if the keywords were less than 30 characters apart. Our computer programs were flexible in setting the window size for \"closeness\". We defined the collocation window as the span of characters around a keyword that are considered \"close\". We ran experiments where the sizes of the collocation windows were set to 10, 20, and 30 characters. A collocation window of 30 characters will consider 30 characters on the left and on the right side of a keyword. The historians observed the computed collocations and preferred the size of 30.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Temporal Analysis of Important Collocations",
                "sec_num": "4."
            },
            {
                "text": "We analyzed the statistics of collocations in the documents about the concept of \"Chinese People\" (\"\u83ef\u4eba,\" hua2 ren2) in Diplomacy (cf. Table 1 ). We identified the keywords with the procedure that we applied to find individual keywords in Constitution that we explained in the previous section. Historians then chose the keywords of interest and we ran Applying Language Technologies to Historical Studies in Chinese the computer programs to do the temporal analysis of the important collocations. This procedure is similar to the procedure that we used to obtain Figure 3 ; the only difference was whether the target of analysis was keywords or collocations. Figure 4 show that the annual percentages of four collocations varied over the years between 1875 and 1909. Significantly large annual percentages again coincided with historical events of the given years. In 1894, the United States of America (USA) (\"\u7f8e\u56fd\" (mei3 guo2) in the chart) and the Qing government signed a treaty to limit Chinese laborers (\"\u534e\u5de5\") entering the USA. 16 In 1905, Chinese societies started to boycott American's products, mostly because the USA would extend the treaty signed in 1894. 17 Initially, the Qing dynasty was trying to protect only the Chinese laborers. Later, the protection was extended to Chinese merchants then extended to Chinese people (Jin et al., 2012) .",
                "cite_spans": [
                    {
                        "start": 1032,
                        "end": 1034,
                        "text": "16",
                        "ref_id": null
                    },
                    {
                        "start": 1165,
                        "end": 1167,
                        "text": "17",
                        "ref_id": null
                    },
                    {
                        "start": 1333,
                        "end": 1351,
                        "text": "(Jin et al., 2012)",
                        "ref_id": "BIBREF10"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 134,
                        "end": 141,
                        "text": "Table 1",
                        "ref_id": "TABREF0"
                    },
                    {
                        "start": 563,
                        "end": 571,
                        "text": "Figure 3",
                        "ref_id": "FIGREF1"
                    },
                    {
                        "start": 659,
                        "end": 667,
                        "text": "Figure 4",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Temporal Analysis of Important Collocations",
                "sec_num": "4."
            },
            {
                "text": "As the statistics in Table 1 show, there can be thousands of documents containing millions of characters in a particular collection. Finding the most relevant documents or essays to read was not easy in the past. With our ability to identify the important keywords and collocations, we could rank the documents based on how documents included the important keywords and 16 \u300a\u4e2d\u7f8e\u83ef\u5de5\u689d\u7d04\u300b\u3001\u300a\u9650\u7981\uf92d\u7f8e\u83ef\u5de5\u4fdd\u8b77\u5bd3\u7f8e\u83ef\u4eba\u689d\u7d04\u300b\uff1a http://dict.zwbk.org/zh-tw/Word_Show/64744.aspx 17 http://zh.wikipedia.org/wiki/\u62b5\u5236\u7f8e\u8ca8\u904b\u52d5\uff1b\u300a\u7c4c\u62d2\u7f8e\u570b\u83ef\u5de5\u7981\u7d04\u516c\u555f\u300b Table 2 shows a part of the table where we ranked the documents in Constitution (cf. Table 1 ). The weights in Table 2 were calculated based on the number of keywords that were used in a document. Larger weights imply that more keywords were used in the document, so the document might be more relevant to the research topic for which the researcher selected the keywords. The ranking function and other techniques for information retrieval and extraction could provide useful information for historians to study specific issues (Jin et al., 2011; Jin et al., 2012) .",
                "cite_spans": [
                    {
                        "start": 370,
                        "end": 372,
                        "text": "16",
                        "ref_id": null
                    },
                    {
                        "start": 1029,
                        "end": 1047,
                        "text": "(Jin et al., 2011;",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 1048,
                        "end": 1065,
                        "text": "Jin et al., 2012)",
                        "ref_id": "BIBREF10"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 21,
                        "end": 28,
                        "text": "Table 1",
                        "ref_id": "TABREF0"
                    },
                    {
                        "start": 500,
                        "end": 507,
                        "text": "Table 2",
                        "ref_id": "TABREF2"
                    },
                    {
                        "start": 585,
                        "end": 592,
                        "text": "Table 1",
                        "ref_id": "TABREF0"
                    },
                    {
                        "start": 611,
                        "end": 618,
                        "text": "Table 2",
                        "ref_id": "TABREF2"
                    }
                ],
                "eq_spans": [],
                "section": "Ranking Individual Documents: An Application of Information Retrieval",
                "sec_num": "5."
            },
            {
                "text": "In this section, we discuss some technical problems related to using computing techniques to support historical studies in Chinese.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Discussion",
                "sec_num": "6."
            },
            {
                "text": "We have illustrated three possible applications of textual analysis for historical studies in previous sections. The applications were based on the frequencies of keywords in the collections. In NLP, we can refer to the frequencies of keywords as term frequencies. In addition, we relied on the \"time stamps\" of the documents, where the \"time stamps\" are the recorded times of the documents. Based on our dependence on the terms frequencies and time stamps, we obtained and presented the figures that we discussed in Sections 3 and 4.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "In these examples, we presume that the frequencies reflect the importance of the concepts that are represented by the terms and collocations, and the results of our work are quite convincing. Nevertheless, we have to watch for the problems of lexical ambiguity and pragmatics that are hidden under the term frequencies.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "For instance, frequently cited events of the past may induce confusion about the significance of term frequencies. Tu et al. (2011) discovered that, although \"\u5f35\u516c\u85dd\" (zhang1 gong1 yi4) was a Chinese name that appeared frequently in collections in the Taiwan History Digital Library (THDL), \"\u5f35\u516c\u85dd\" referred to a person who actually lived in the Tang dynasty 18 which is well before the time period of the documents in THDL. The documents in THDL referred to\"\u5f35\u516c\u85dd\" because of a story that was well-known in the Qing dynasty (1644AD-1912AD). That the term frequency of \"\u5f35\u516c\u85dd\" is high in THDL does not imply that \"\u5f35\u516c\u85dd\" himself was an important person in Taiwan in the Qing dynasty.",
                "cite_spans": [
                    {
                        "start": 115,
                        "end": 131,
                        "text": "Tu et al. (2011)",
                        "ref_id": "BIBREF15"
                    },
                    {
                        "start": 354,
                        "end": 356,
                        "text": "18",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "(618AD-1907AD),",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "Lexical ambiguity may make the term frequencies less reliable. Yu (2012) accentuates this issue with \"\u6c11\u4e3b\" (min2 zhu3). In modern Chinese, \"\u6c11\u4e3b\" is the word for \"democracy\". Nevertheless, it could represent the emperor (\u6c11\u4e4b\u4e3b, min2 zhi1 zhu3), the American president, and the Republic (in \u6c11\u4e3b\u570b, min2 zhu3 guo2) in non-modern Chinese text.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "The first author of this paper examined the DGOROC, and found that \"\u9673\u5efa\u4e2d\" was a very common name in the database. Hence, finding the actual identities of names is an important issue, in addition to computing the term frequencies. Distinguishing persons of the same name in modern databases requires extraordinary sources of private information.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "Although differentiating persons with the same names is not easy, identifying names in Chinese text is not an easy task for the research of Named Entity Recognition (often referred to as NER (Wu et al., 2006) ) in the first place. For instance, it may not be easy to extract names from Chinese text like \"\u4e2d\u592e\u9ad8\u5c64\u6b63\u919e\u91c0\u5b89\u6392\uf9a8\u8a08\u756b\u63a5\u4efb\u4e2d\u7d44\u90e8\u9577\" 19 (zhong1 yang1 gao1 ceng2 zheng4 yun4 niang4 an1 pai2 ling4 ji4 hua4 jie1 ren4 zhong1 zu3 bu4 zhang3) if we do not know \"\uf9a8\u8a08\u5283\" (ling4 ji4 hua4) is a name. 20",
                "cite_spans": [
                    {
                        "start": 191,
                        "end": 208,
                        "text": "(Wu et al., 2006)",
                        "ref_id": "BIBREF16"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Lexical Ambiguity, Pragmatics, and Term Identity",
                "sec_num": "6.1"
            },
            {
                "text": "In Section 6.1, we discuss the interpretation of a given term. In Chinese, however, we also have to define the concept of \"term\". If we cannot define terms precisely, then we have no grounds for defining collocations. It is known that Chinese words are not separated by spaces like in alphabetic languages, and the task of separating Chinese words in Chinese text is generally called word segmentation (e.g., Ma & Chen, 2005; Jiang et al., 2006; Tseng et al., 2005) . It is less known, however, that pre-modern Chinese text does not have punctuation, and readers also have to figure out the divisions of sentences (Huang, 2008) .",
                "cite_spans": [
                    {
                        "start": 409,
                        "end": 425,
                        "text": "Ma & Chen, 2005;",
                        "ref_id": "BIBREF13"
                    },
                    {
                        "start": 426,
                        "end": 445,
                        "text": "Jiang et al., 2006;",
                        "ref_id": "BIBREF9"
                    },
                    {
                        "start": 446,
                        "end": 465,
                        "text": "Tseng et al., 2005)",
                        "ref_id": "BIBREF14"
                    },
                    {
                        "start": 614,
                        "end": 627,
                        "text": "(Huang, 2008)",
                        "ref_id": "BIBREF8"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Word Segmentation and Sentence Division",
                "sec_num": "6.2"
            },
            {
                "text": "Clearly, if we could not divide sentences and segment words correctly, we would not be able to acquire correct term frequencies. This may happen when we process text like \"\u4e94\ufa08\u8005 \uf90a\u4e3b\u7fa9\u6728\u4e3b\u4ec1\u6c34\u4e3b\u667a\u706b\u4e3b\uf9b6\u571f\u4e3b\u4fe1\" (wu3 xing2 zhe3 jin1 zhu3 yi4 mu4 zhu3 ren2 shui3 zhu3 zhi4 huo3 zhu3 li3 tu3 zhu3 xin4). We would have to add punctuation to divide this string: \"\u4e94\ufa08\u8005\uff0c\uf90a\u4e3b\u7fa9\uff0c\u6728\u4e3b\u4ec1\uff0c\u6c34\u4e3b\u667a\uff0c\u706b\u4e3b\uf9b6\uff0c\u571f\u4e3b\u4fe1\". After this step, we know that \"\u667a\u706b\" is not a term in the original string, although \"\u667a\u706b\" can be a meaningful term in modern Chinese. 21 Given the divided string, we still have to face the word segmentation problem. In this example, each character in \"\uf90a\u4e3b\u7fa9\" represents a specific meaning. We cannot interpret \"\u4e3b\u7fa9\" in \"\uf90a\u4e3b\u7fa9\" as we would interpret \"\u4e3b\u7fa9\" (-ism) in \"\u5e1d\u570b\u4e3b\u7fa9\" (di4 guo2 zhu3 yi4; imperialism) or \"\u8cc7\u672c\u4e3b\u7fa9\" (zi1 ben3 zhu3 yi4; capitalism) in modern Chinese. Similarly, we have to know that \"\uf90a\u4e3b\" is not a term in the original string, although \"\uf90a\u4e3b\" (a wealthy person) is a meaningful term in modern Chinese.",
                "cite_spans": [
                    {
                        "start": 497,
                        "end": 499,
                        "text": "21",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Word Segmentation and Sentence Division",
                "sec_num": "6.2"
            },
            {
                "text": "An actual problem took place when we used the DSMCTL to investigate whether energy conservation was a concern in the Qing dynasty (Chou et al., 2012) . Without a Chinese segmenter for pre-modern Chinese, we found many occurrences of \"\u80fd\u6e90\" (neng2 yuan2) in the database, but, most of the time, \"\u80fd\u6e90\" was just a sub-string of \"\uf967\u80fd\u6e90\u6e90\u800c\uf92d\" (bu4 neng2 yuan2 yuan2 er2 lai2) when people talked about something that could not come indefinitely. 22",
                "cite_spans": [
                    {
                        "start": 130,
                        "end": 149,
                        "text": "(Chou et al., 2012)",
                        "ref_id": "BIBREF4"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Word Segmentation and Sentence Division",
                "sec_num": "6.2"
            },
            {
                "text": "In Sections 3 and 4, we briefly introduced applications of temporal trends of keywords ( Figure  3 ) and trends of collocations (Figure 4 ) that were more thoroughly discussed in Jin et al. (2011) and Jin et al. (2012) , respectively. Researchers in other fields also have found impressive applications of trends of keywords (e.g., Caneior & Mylonakis, 2009) . Despite these successful applications, caution is in need to interpret the observed trends. Figure 5 shows temporal trends for the names of three main characters in a famous novel Dream of the Red Chamber (DRC). The horizontal axis shows the chapters of the DRC. The vertical axis shows the frequency of the keywords (persons' names in this chart). The highs of the curves shows the times of being mentioned of a person in a particular chapter, so are indicative of the relatively importance of the persons. We discuss three main persons in DRC, \"\u5bf6\u7389\" (Bao3 Yu4), \"\u9edb\u7389\" (Dai4 Yu4), and \"\u5bf6\u91f5\" (Bao3 Tsai1), in the following. 21 \"\u667a\u706b\" happens to be the name of a Chinese company: http://www.zhihuo.asia/. 22 The Chinese segmentation service at Academia Sinica (http://ckipsvr.iis.sinica.edu.tw/) would return \"\uf967\u80fd,\" \"\u6e90\u6e90,\" \"\u800c,\" and \"\uf92d\" for \"\uf967\u80fd\u6e90\u6e90\u800c\uf92d\". The online version of the Stanford parser (http://nlp.stanford.edu:8080/parser/index.jsp) would return \"\uf967,\" \"\u80fd,\" and \"\u6e90\u6e90\u800c\uf92d\".",
                "cite_spans": [
                    {
                        "start": 179,
                        "end": 196,
                        "text": "Jin et al. (2011)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 201,
                        "end": 218,
                        "text": "Jin et al. (2012)",
                        "ref_id": "BIBREF10"
                    },
                    {
                        "start": 332,
                        "end": 358,
                        "text": "Caneior & Mylonakis, 2009)",
                        "ref_id": null
                    },
                    {
                        "start": 982,
                        "end": 984,
                        "text": "21",
                        "ref_id": null
                    },
                    {
                        "start": 1060,
                        "end": 1062,
                        "text": "22",
                        "ref_id": null
                    }
                ],
                "ref_spans": [
                    {
                        "start": 89,
                        "end": 98,
                        "text": "Figure  3",
                        "ref_id": "FIGREF1"
                    },
                    {
                        "start": 128,
                        "end": 137,
                        "text": "(Figure 4",
                        "ref_id": null
                    },
                    {
                        "start": 453,
                        "end": 461,
                        "text": "Figure 5",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Trends: Informative or Deceptive",
                "sec_num": "6.3"
            },
            {
                "text": "Do the ups and downs of a particular curve show the changes of importance of a person?",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Some Chances and Challenges in 39 Applying Language Technologies to Historical Studies in Chinese",
                "sec_num": null
            },
            {
                "text": "Intuitively, the answer may be yes. If the name of a person was mentioned more frequently, that particular person should be more involved in a chapter. This interpretation, however, is not flawless -a person being mentioned more times might be the result of a longer chapter. Being mentioned more times in a longer chapter might not be solid proof of the importance of the mentioned person. Figure 6 shows the numbers of characters in each chapter in DRC. Evidently, some chapters are longer and some are shorter. 0   2000   4000   6000   8000   10000   12000   14000   1  6  11  16  21  26  31  36  41  46  51  56  61  66  71  76  81  86  91  96  101  106  111 Let and , respectively, denote the frequency of a keyword and the length of a chapter in DRC. We divide by , for =1, \u2026, 120, for the three names in Figure 5 . Figure 7 shows the resulting curves for the three persons.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 391,
                        "end": 399,
                        "text": "Figure 6",
                        "ref_id": null
                    },
                    {
                        "start": 514,
                        "end": 661,
                        "text": "0   2000   4000   6000   8000   10000   12000   14000   1  6  11  16  21  26  31  36  41  46  51  56  61  66  71  76  81  86  91  96  101  106  111",
                        "ref_id": "TABREF0"
                    },
                    {
                        "start": 810,
                        "end": 818,
                        "text": "Figure 5",
                        "ref_id": null
                    },
                    {
                        "start": 821,
                        "end": 829,
                        "text": "Figure 7",
                        "ref_id": "FIGREF6"
                    }
                ],
                "eq_spans": [],
                "section": "Some Chances and Challenges in 39 Applying Language Technologies to Historical Studies in Chinese",
                "sec_num": null
            },
            {
                "text": "We can observe some important changes in the curves. Take the curve for Bao-Yu (\"\u5bf6 \u7389\") for example. Bao-Yu was mentioned 84, 116, and 98 times in Chapters 8, 19, and 28, respectively. These three instances formed the first three peaks above 80 in the curve for Bao-Yu in Figure 5 . The frequencies may have suggested that Bao-Yu were more important in Chapter 19 than in Chapters 8 and 28. After we divided these frequencies by the chapter lengths, we observed that the proportions of Bao-Yu being mentioned in these chapters were almost the same in Figure 7 . Hence, the trends illustrated in Figures 6 and 7 provide hints for different conclusions.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 271,
                        "end": 279,
                        "text": "Figure 5",
                        "ref_id": null
                    },
                    {
                        "start": 550,
                        "end": 558,
                        "text": "Figure 7",
                        "ref_id": "FIGREF6"
                    },
                    {
                        "start": 594,
                        "end": 609,
                        "text": "Figures 6 and 7",
                        "ref_id": "FIGREF6"
                    }
                ],
                "eq_spans": [],
                "section": "Figure 6. Lengths of chapters in Dream of the Red Chamber",
                "sec_num": null
            },
            {
                "text": "Consider another example. Assume that we want to know who among the three persons liked to \"smile and say\" most in DRC. Curves in Figure 8 show the frequencies of \"\u5bf6\u7389\u7b11\u9053,\" \"\u9edb\u7389\u7b11\u9053,\" and \"\u5bf6\u91f5\u7b11\u9053\" in DRC, where \"\u7b11\u9053\" (xiao4 dao4) is a way to say \"smile and say\". The curves suggest that, before Chapter 40, Bao-Yu was the person who liked to \"smile and say\" most.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 130,
                        "end": 138,
                        "text": "Figure 8",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Figure 6. Lengths of chapters in Dream of the Red Chamber",
                "sec_num": null
            },
            {
                "text": "Nevertheless, one may contend that the absolute frequency may not be a perfect indicator for how likely a person was to \"smile and say\". If a person was mentioned less frequently, then s/he would not be able to \"smile and say\" as frequently as another who was mentioned more frequently. Let and , respectively, denote the frequency a person \"smiled and said\" and a person was mentioned in a chapter . For the three persons in our current discussion, was their individual term frequency that we showed in Figure 5 . We divided by for each person and came up with Figure 9 .",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 504,
                        "end": 512,
                        "text": "Figure 5",
                        "ref_id": null
                    },
                    {
                        "start": 562,
                        "end": 570,
                        "text": "Figure 9",
                        "ref_id": "FIGREF7"
                    }
                ],
                "eq_spans": [],
                "section": "Figure 6. Lengths of chapters in Dream of the Red Chamber",
                "sec_num": null
            },
            {
                "text": "Quite interestingly, the curve for Bao-Yu does not dominate the others anymore. Instead, Bao-Tsai (\"\u5bf6\u91f5\") smiled and said something once every two appearances in Chapter 19, as did Dai-Yu (\"\u9edb\u7389\") in Chapter 73. In fact, never did Bao-Yu smile and say as often as 50% of the time he appeared in any chapter. The highest proportion of Bao-Yu's \"smile and say\" took place in Chapter 88, where the proportion still fell short of 40%.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Figure 6. Lengths of chapters in Dream of the Red Chamber",
                "sec_num": null
            },
            {
                "text": "One researcher may be interested in the times a person smiled and said something, and another might be interested in the proportion a person smiled and said something when the person was mentioned in DRC. Take Bao-Yu for example. In the former case (Figure 8 ), the term frequency of Bao-Yu is the focus. In contrast, in the latter case (Figure 9 ), the conditional probability Pr Bao-Yu-Xiao-Dao Bao-Yu is of interest, and we have to compute the probability based on the observed frequencies. Different trends and analyses should be used for different purposes, and this is up to the researchers' discretion. When designing tools for assisting historical studies, appropriate functionalities should be considered and explained to their users as clear as possible.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 249,
                        "end": 258,
                        "text": "(Figure 8",
                        "ref_id": null
                    },
                    {
                        "start": 337,
                        "end": 346,
                        "text": "(Figure 9",
                        "ref_id": "FIGREF7"
                    }
                ],
                "eq_spans": [],
                "section": "Figure 6. Lengths of chapters in Dream of the Red Chamber",
                "sec_num": null
            },
            {
                "text": "In addition to the ability to process normal pre-modern Chinese text, one may need to handle transliterated and translated words. Chinese people encountered western culture more directly and more frequently starting from late 1500s. Transliteration and translation are important ways for people to use Chinese words to convey and understand western concepts and entities.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Transliteration and Translation",
                "sec_num": "6.4"
            },
            {
                "text": "To study the interactions between the Chinese and western cultures in pre-modern times, getting to know the Chinese transliterations and translations is an important step. For instance, \"president\" was transliterated into \"\u4f2f\uf9e4\u74bd\" (bo2 li3 si2), \"\u4f2f\uf9e4\u559c\u9813\" (bo2 li3 si2 dun4), \"\u4f2f \uf9e4\u74bd\u5929\u5fb7\" (bo2 li3 si2 tian1 de2), and \"\u4f2f\uf98a\u932b\u5929\u5fb7\" (bo2 li3 si2 tian1 de2). Some Chinese characters were selected based on the pronunciations of \"president\", and some were selected to show respect to the position of \"president.\" \"Pacific Ocean\" was translated into \"\u5927\u6d77\" (da4 hai3), \"\u5927\u6771\u6d0b\" (da4 dong1 yang1), and \"\u592a\u6d0b\u6d77\" (tai4 yang1 hai3), and transliterated into \"\u5351\u897f\u6ea2\u6e56\" (bei4 si1 yi1 hu2) and \"\u6bd4\u897f\u975e\u76ca\u6d77\" (bi3 si1 fei1 yi4 hai3). The translations show people's knowledge about the size and position of the Pacific Ocean. \"Politics\" was transliterated into \"\u8584\uf9dd\u7b2c\u52a0\" (bo2 li4 di4 jia1) and \"\u6ce2\uf9f7\u7279\" (po2 li4 te4); both were transliterations.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Transliteration and Translation",
                "sec_num": "6.4"
            },
            {
                "text": "Historians may spend their lifetimes identifying the translated and transliterated terms in historical documents. If one could provide researchers the Chinese terms for the western concepts and entities, the researchers would be able to investigate and understand how Chinese faced the West hundreds years ago.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Transliteration and Translation",
                "sec_num": "6.4"
            },
            {
                "text": "Therefore, we imagine that it would be useful if computing technologies could help historical researchers identify transliterated and translated terms in historical documents. It may be not easy to use human experts to annotate a database that has 120 million characters, ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Transliteration and Translation",
                "sec_num": "6.4"
            },
            {
                "text": "An anonymous reviewer of the manuscript of this paper pointed out that applications of advanced NLP techniques will strengthen the values of the collected statistics. For instance, one may classify the keywords into types and conduct temporal analysis on keywords of the same type. This may give us a trend analysis similar to the analysis of emotion trend reported in Yang et al. (2007) . It is also possible to treat the network of keywords as a social network. The nodes can be verbs, nouns, and names, and links show strengths of associativity. Networks like this may shed light on historical events that were difficult to see by simply studying the historical documents.",
                "cite_spans": [
                    {
                        "start": 369,
                        "end": 387,
                        "text": "Yang et al. (2007)",
                        "ref_id": "BIBREF17"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Advanced NLP Techniques: Trend Analysis",
                "sec_num": "6.5"
            },
            {
                "text": "The DGOROC database 23 provides information about the appointments of government officials of the Republic of China in Taiwan. This database was constructed and verified with human labor. Information was copied from hard copies of official documents, entered into text files, and was verified for quality assurance. It contains more than 850 thousand records dated from 1912, and is useful for studying modern history and relevant applications about Taiwan and China (e.g., Liu & Lai, 2011) .",
                "cite_spans": [
                    {
                        "start": 474,
                        "end": 490,
                        "text": "Liu & Lai, 2011)",
                        "ref_id": "BIBREF12"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "Since the data came from a real and changing government, there can be barriers that were difficult to overcome simply by computing technology. For instance, the current government in Taiwan was not in a really stable condition until she moved to Taiwan in 1949. Hence, the database is relatively more complete for records after circa 1949.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "It should not be surprising that a government tries and evolves to serve the nation in the fast changing world. For instance, there was no \"Ministry of Education\" before 1928, although there must have been some government agents to handle national education policies before 1928. Hence, a researcher will have to know the names of the agents that were responsible for education to study the national education policies circa 1928. In this case, a simple keyword search service may not help very much.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "Although the data collected after 1949 was more complete in DGOROC, the government may change the rules for whether or not to announce some types of assignments. For instance, there are departments in the Ministry of Education, and the department heads may change their appointments from a department to another, but this type of switch is not publicly announced in recent years.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "The appointments of lower ranks of government officials may not be announced at real time. The announcement of such appointments may be delayed so that a larger group of appointments would be announced at the same time. If the time stamps of events for a certain study matter, then this kind of delay may be troublesome.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "Despite these remaining challenges in DGOROC, we consider this database unique and important. By incorporating information available from other database maintenance agents of the central government and from national libraries, the database will offer researchers a great information source for studying modern history of Taiwan.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Time Stamps, Missing Data, and Fundamental Changes in DGOROC",
                "sec_num": "6.6"
            },
            {
                "text": "We delineate our experience in using three sources of historical documents in Chinese: the database of Chinese historical documents that contain more than 120 million simplified Chinese characters, the Database of Government Officials of the Republic of China, and the Dream of the Red Chamber. Techniques for natural language processing were employed to analyze the contents of the documents to facilitate the studies of historical events. The exploration showed that NLP techniques are instrumental for the studies of non-modern Chinese historical documents. Our experience also suggested that advanced NLP techniques and more complete data collection are necessary for supporting research work in more precise ways.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Concluding Remarks",
                "sec_num": "7."
            },
            {
                "text": "\uf9f7\u5baa: li4 xian4, constitutionality 13 \u9009\u4e3e: xuan3 ju3, election; \u7ae0\u7a0b: zhang1 cheng2, rules",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u7b79\u5907: chou2 bei4, preparation; \u81ea\u6cbb: zi4 zhi4, self-governance 15 http://www.google.com/trends/",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "http://en.wikipedia.org/wiki/Tang_Dynasty 19 Source: http://www.cbfcn.com/news_detail.aspx?strnew=1154 20 In fact, \"\uf9a8\uf937\u7dda\" (ling4 lu4 xian4), \"\uf9a8\u653f\u7b56\" (ling4 zheng4 ce4), \"\uf9a8\u5b8c\u6210\" (ling4 wan2 cheng2), and \" \uf9a8 \u65b9 \u91dd \" (ling4 fang1 jhen1) are also Chinese names (although they are quite unusual) : http://zh.wikipedia.org/wiki/\uf9a8\u8a08\u5283",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "The first author gained experience with DGOROC while serving as the project leader for maintaining DGOROC between February and August 2011. The comments about DGOROC in this section are of the first author.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [
            {
                "text": "The work was supported in part by the funding from the National Science Council in Taiwan under the contracts NSC-99-2221-E-004-007 and NSC-100-2221-E-004-014. The Database for the Study of Modern Chinese Thought and Literature was funded by the sub-project 100H51 at the National Chengchi University, Aim for the Top Universities Project, Ministry of Education, Taiwan. The writing of this paper benefits tremendously from reviewers' comments on previous versions of this paper. Section 6.5, in particular, was prompted by the review comments of this journal version. The English editor, Mr. Sun, of this journal helped us a lot by providing suggestions for better presentation in addition to correcting language problems.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Acknowledgments",
                "sec_num": null
            }
        ],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Google trends: A Web-based tool for realtime surveillance of disease outbreaks",
                "authors": [
                    {
                        "first": "H",
                        "middle": [
                            "A"
                        ],
                        "last": "Carneior",
                        "suffix": ""
                    },
                    {
                        "first": "E",
                        "middle": [],
                        "last": "Mylonakis",
                        "suffix": ""
                    }
                ],
                "year": 2009,
                "venue": "Clinical Infectious Diseases",
                "volume": "49",
                "issue": "10",
                "pages": "1557--1564",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Carneior, H. A. & Mylonakis, E. (2009). Google trends: A Web-based tool for realtime surveillance of disease outbreaks. Clinical Infectious Diseases, 49(10), 1557-1564.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Some Chances and Challenges in 45",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Some Chances and Challenges in 45",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Applying Language Technologies to Historical Studies in Chinese",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Applying Language Technologies to Historical Studies in Chinese",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "PAT-tree-based adaptive keyphrase extraction for intelligent Chinese information retrieval",
                "authors": [
                    {
                        "first": "L.-F",
                        "middle": [],
                        "last": "Chien",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Information Processing and Management",
                "volume": "35",
                "issue": "4",
                "pages": "501--521",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chien, L.-F.. (1999). PAT-tree-based adaptive keyphrase extraction for intelligent Chinese information retrieval. Information Processing and Management, 35(4), 501-521.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "\u5f9e\uf98c\u53f2\u6587\u737b\u95dc\u9375\u5b57\uf92d\u770b--\u6e05\u672b\u6c11\u521d\u767e\uf98e\u9593\u4e2d\u570b \u80fd\u6e90\u89c0\uf9a3\u6f14\u9032\u521d\u63a2, Energy Monthly, to appear",
                "authors": [
                    {
                        "first": "L.-F",
                        "middle": [],
                        "last": "Chou",
                        "suffix": ""
                    },
                    {
                        "first": "C.-L",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "Y.-S",
                        "middle": [],
                        "last": "Yu",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chou, L.-F., Liu, C.-L., & Yu, Y.-S. (2012). \u5f9e\uf98c\u53f2\u6587\u737b\u95dc\u9375\u5b57\uf92d\u770b--\u6e05\u672b\u6c11\u521d\u767e\uf98e\u9593\u4e2d\u570b \u80fd\u6e90\u89c0\uf9a3\u6f14\u9032\u521d\u63a2, Energy Monthly, to appear, Taiwan Institute of Economic Research. (in Chinese)",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Extension of Zipf's law to word and character n-grams for English and Chinese",
                "authors": [
                    {
                        "first": "L",
                        "middle": [
                            "Q"
                        ],
                        "last": "Ha",
                        "suffix": ""
                    },
                    {
                        "first": "E",
                        "middle": [
                            "I"
                        ],
                        "last": "Sicilia-Garcia",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Ming",
                        "suffix": ""
                    },
                    {
                        "first": "F",
                        "middle": [
                            "J"
                        ],
                        "last": "Smith",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "International Journal of Computational Linguistics and Chinese Language Processing",
                "volume": "8",
                "issue": "1",
                "pages": "77--102",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Ha, L. Q., Sicilia-Garcia, E. I., Ming, J., & Smith, F. J. (2003). Extension of Zipf's law to word and character n-grams for English and Chinese. International Journal of Computational Linguistics and Chinese Language Processing, 8(1), 77-102.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "From Preservation to Knowledge Creation",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Hsiang",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "The Way to Digital Humanities (\u5f9e\u4fdd\u5b58\u5230\u5275\u9020\uff1a\u958b\u555f\uf969\u4f4d\u4eba\u6587\u7814\u7a76), Series on Digital Humanities",
                "volume": "1",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Hsiang, J. (Ed.) (2011a). From Preservation to Knowledge Creation: The Way to Digital Humanities (\u5f9e\u4fdd\u5b58\u5230\u5275\u9020\uff1a\u958b\u555f\uf969\u4f4d\u4eba\u6587\u7814\u7a76), Series on Digital Humanities, volume 1, National Taiwan University Press. (in Chinese)",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "New Eyes for Discovery: Foundations and Imaginations of Digital Humanities (\uf969\u4f4d\u4eba\u6587\u7814\u7a76\u7684\u65b0\u8996\u91ce\uff1a\u57fa\u790e\u8207\u60f3\u50cf), Series on Digital Humanities",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Hsiang",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Ed",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "2",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Hsiang, J., (Ed.) (2011b). New Eyes for Discovery: Foundations and Imaginations of Digital Humanities (\uf969\u4f4d\u4eba\u6587\u7814\u7a76\u7684\u65b0\u8996\u91ce\uff1a\u57fa\u790e\u8207\u60f3\u50cf), Series on Digital Humanities, volume 2, National Taiwan University Press. (in Chinese)",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "Classical Chinese Sentence Division by Sequence Labeling Approaches (\u4ee5\u5e8f\uf99c \u6a19\u8a18\u65b9\u6cd5\u89e3\u6c7a\u53e4\u6f22\u8a9e\u65b7\uf906\u554f\u984c)",
                "authors": [
                    {
                        "first": "H.-H",
                        "middle": [],
                        "last": "Huang",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Huang, H.-H. (2008). Classical Chinese Sentence Division by Sequence Labeling Approaches (\u4ee5\u5e8f\uf99c \u6a19\u8a18\u65b9\u6cd5\u89e3\u6c7a\u53e4\u6f22\u8a9e\u65b7\uf906\u554f\u984c), Master's Thesis, National Chiao Tung University, Hsinchu, Taiwan. (in Chinese)",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "A pragmatic Chinese word segmentation approach based on mixing models",
                "authors": [
                    {
                        "first": "W",
                        "middle": [],
                        "last": "Jiang",
                        "suffix": ""
                    },
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Guan",
                        "suffix": ""
                    },
                    {
                        "first": "X.-L",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "International Journal of Computational Linguistics and Chinese Language Processing",
                "volume": "11",
                "issue": "4",
                "pages": "393--416",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jiang, W., Guan, Y., & Wang, X.-L. (2006). A pragmatic Chinese word segmentation approach based on mixing models, International Journal of Computational Linguistics and Chinese Language Processing, 11(4), 393-416.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Frequency analysis and application of 'co-occurrence' phrases: the origin of the concept 'Hua-Ren' as an example",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Jin",
                        "suffix": ""
                    },
                    {
                        "first": "W.-Y",
                        "middle": [],
                        "last": "Chiu",
                        "suffix": ""
                    },
                    {
                        "first": "C.-L",
                        "middle": [
                            ";"
                        ],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "\u300c\u5171\u73fe\u300d\u8a5e \u983b \u5206 \u6790 \u53ca \u5176\u904b \u7528 -\u4ee5 \u300c\u83ef \u4eba \u300d \u89c0 \uf9a3\u8d77 \u6e90 \u70ba \uf9b5",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "Series on Digital Humanities",
                "volume": "3",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jin, G., Chiu, W.-Y., & Liu, C.-L. (2012). Frequency analysis and application of 'co-occurrence' phrases: the origin of the concept 'Hua-Ren' as an example (\u300c\u5171\u73fe\u300d\u8a5e \u983b \u5206 \u6790 \u53ca \u5176\u904b \u7528 -\u4ee5 \u300c\u83ef \u4eba \u300d \u89c0 \uf9a3\u8d77 \u6e90 \u70ba \uf9b5 ), to appear, in Series on Digital Humanities, volume 3, J. Hsiang (Ed.), National Taiwan University Press. (in Chinese)",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Applications of digital methods to study social movements -Using the preparation of constitutional monarchy in the late Qing dynasty as an example (\u793e\u6703\ufa08\u52d5\u7684\uf969\u4f4d\u4eba\u6587\u7814\u7a76\uff1a\u4ee5\u6e05\u672b\u9810\u5099\uf9f7\u61b2\u70ba\uf9b5)",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Jin",
                        "suffix": ""
                    },
                    {
                        "first": "T.-S",
                        "middle": [],
                        "last": "Yu",
                        "suffix": ""
                    },
                    {
                        "first": "C.-L",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jin, G., Yu, T.-S., & Liu, C.-L. (2011). Applications of digital methods to study social movements -Using the preparation of constitutional monarchy in the late Qing dynasty as an example (\u793e\u6703\ufa08\u52d5\u7684\uf969\u4f4d\u4eba\u6587\u7814\u7a76\uff1a\u4ee5\u6e05\u672b\u9810\u5099\uf9f7\u61b2\u70ba\uf9b5), presented in the Third Conference of Digital Archives and Digital Humanities. (in Chinese)",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "Rank promotion prediction on Taiwanese government officials",
                "authors": [
                    {
                        "first": "J.-S",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "L.-P",
                        "middle": [],
                        "last": "Lai",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "113--130",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Liu, J.-S., & Lai, L.-P. (2011). Rank promotion prediction on Taiwanese government officials, in (Hsiang, 2011a), 113-130. (in Chinese)",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "Design of CKIP Chinese word segmentation system",
                "authors": [
                    {
                        "first": "W.-Y",
                        "middle": [],
                        "last": "Ma",
                        "suffix": ""
                    },
                    {
                        "first": "K.-J",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Chinese and Oriental Languages Information Processing Society",
                "volume": "14",
                "issue": "3",
                "pages": "235--249",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Ma, W.-Y. & Chen, K.-J. (2005). Design of CKIP Chinese word segmentation system. Chinese and Oriental Languages Information Processing Society, 14(3), 235-249.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "A conditional random field word segmenter",
                "authors": [
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Tseng",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Chang",
                        "suffix": ""
                    },
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Andrew",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Jurafsky",
                        "suffix": ""
                    },
                    {
                        "first": "C",
                        "middle": [],
                        "last": "Manning",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proceedings of the Fourth SIGHAN Workshop on Chinese Language Processing",
                "volume": "",
                "issue": "",
                "pages": "32--39",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Tseng, H., Chang, P., Andrew, G., Jurafsky, D., & Manning, C. (2005). A conditional random field word segmenter. Proceedings of the Fourth SIGHAN Workshop on Chinese Language Processing, 32-39.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "Information technology and open problems in the Taiwan history digital library (THDL), in (Hsiang, 2011b)",
                "authors": [
                    {
                        "first": "F.-E",
                        "middle": [],
                        "last": "Tu",
                        "suffix": ""
                    },
                    {
                        "first": "H.-C",
                        "middle": [],
                        "last": "Tu",
                        "suffix": ""
                    },
                    {
                        "first": "S.-P",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "H.-I",
                        "middle": [],
                        "last": "Ho",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Hsiang",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "21--44",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Tu, F.-E., Tu, H.-C., Chen, S.-P., Ho, H.-I., & Hsiang, J. (2011). Information technology and open problems in the Taiwan history digital library (THDL), in (Hsiang, 2011b), 21-44. (in Chinese)",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "On using ensemble methods for Chinese named entity recognition",
                "authors": [
                    {
                        "first": "C.-W",
                        "middle": [],
                        "last": "Wu",
                        "suffix": ""
                    },
                    {
                        "first": "S.-Y",
                        "middle": [],
                        "last": "Jan",
                        "suffix": ""
                    },
                    {
                        "first": "R. T-H",
                        "middle": [],
                        "last": "Tsai",
                        "suffix": ""
                    },
                    {
                        "first": "W.-L",
                        "middle": [],
                        "last": "Hsu",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Proceedings of the Fifth SIGHAN Workshop on Chinese Language Processing",
                "volume": "",
                "issue": "",
                "pages": "142--145",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Wu, C.-W., Jan, S.-Y., Tsai, R. T-H., & Hsu, W.-L. (2006). On using ensemble methods for Chinese named entity recognition. Proceedings of the Fifth SIGHAN Workshop on Chinese Language Processing, 142-145.",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Emotion trend analysis using blog Corpora",
                "authors": [
                    {
                        "first": "C.-H",
                        "middle": [],
                        "last": "Yang",
                        "suffix": ""
                    },
                    {
                        "first": "H.-A",
                        "middle": [],
                        "last": "Kuo",
                        "suffix": ""
                    },
                    {
                        "first": "H.-H",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proceedings of the Nineteenth Conference on Computational Linguistics and Speech Processing",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yang, C.-H., Kuo, H.-A., & Chen, H.-H. (2007). Emotion trend analysis using blog Corpora, Proceedings of the Nineteenth Conference on Computational Linguistics and Speech Processing, [http://aclweb.org/anthology-new/O/O07/O07-1015.pdf]. (in Chinese)",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "\u9c7c\u5b8f\uf977)\u3002(2012)\u3002\u8303\u5f0f\u7684\u8f6c\u53d8\uff1a\u91cd\u5efa\u89c2\uf9a3\u53f2\u56fe\u50cf\u4e2d\u7684\u5386\u53f2\u771f\u5b9e -\u65b0\u65b9\u6cd5\u4e0e\u4e2d\u56fd \u8fd1\u4ee3\u89c2\uf9a3\u53f2\u7814\u7a76\uff0c\u6771\u4e9e\u89c0\uf9a3\u53f2\u96c6\u520a\uff0c\u7b2c\u4e09\u671f\uff0c\u5c07\u51fa\u520a\u3002\u653f\u6cbb\u5927\u5b78\u51fa\u7248\u793e\u3002",
                "authors": [
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Yu",
                        "suffix": ""
                    }
                ],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yu, H. (\u9c7c\u5b8f\uf977)\u3002(2012)\u3002\u8303\u5f0f\u7684\u8f6c\u53d8\uff1a\u91cd\u5efa\u89c2\uf9a3\u53f2\u56fe\u50cf\u4e2d\u7684\u5386\u53f2\u771f\u5b9e -\u65b0\u65b9\u6cd5\u4e0e\u4e2d\u56fd \u8fd1\u4ee3\u89c2\uf9a3\u53f2\u7814\u7a76\uff0c\u6771\u4e9e\u89c0\uf9a3\u53f2\u96c6\u520a\uff0c\u7b2c\u4e09\u671f\uff0c\u5c07\u51fa\u520a\u3002\u653f\u6cbb\u5927\u5b78\u51fa\u7248\u793e\u3002(in Chinese)",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Connecting text mining and natural language processing in a humanistic context",
                "authors": [
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Xiang",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Unsworth",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Proceedings of the International Conference on Digital Humanities",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Xiang, X., & Unsworth, J. (2006). Connecting text mining and natural language processing in a humanistic context, Proceedings of the International Conference on Digital Humanities 2006.",
                "links": null
            },
            "BIBREF20": {
                "ref_id": "b20",
                "title": "On the applicability of Zipf's law in Chinese word frequency distribution",
                "authors": [
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Xiao",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "This journal has been renamed as International Journal of Asian Language Processing",
                "volume": "18",
                "issue": "",
                "pages": "33--46",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Xiao, H.. (2008). On the applicability of Zipf's law in Chinese word frequency distribution, Journal of Chinese Language and Computing, 18(1), 33-46. (This journal has been renamed as International Journal of Asian Language Processing in 2009.)",
                "links": null
            },
            "BIBREF21": {
                "ref_id": "b21",
                "title": "Human Behavior and the Principle of Least Effort",
                "authors": [
                    {
                        "first": "G",
                        "middle": [
                            "K"
                        ],
                        "last": "Zipf",
                        "suffix": ""
                    }
                ],
                "year": 1949,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Zipf, G. K.. (1949). Human Behavior and the Principle of Least Effort. Addison Wesley.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "uris": null,
                "type_str": "figure",
                "text": "Pseudowords in the Chinese historical collections abide by Zipf'",
                "num": null
            },
            "FIGREF1": {
                "uris": null,
                "type_str": "figure",
                "text": "shows the curves of annual percentages of all words (Total) and six keywords over the years between 1905 and 1911 in Constitution (cf.Table 1).",
                "num": null
            },
            "FIGREF2": {
                "uris": null,
                "type_str": "figure",
                "text": "Figure 2. Reducing the influences of sizes of individual collections",
                "num": null
            },
            "FIGREF3": {
                "uris": null,
                "type_str": "figure",
                "text": "14 became relatively more important.",
                "num": null
            },
            "FIGREF4": {
                "uris": null,
                "type_str": "figure",
                "text": "Figure 4. Importance of keyword collocations varied over the years",
                "num": null
            },
            "FIGREF5": {
                "uris": null,
                "type_str": "figure",
                "text": "Figure 5. Frequencies of three main names in Dream of the Red Chamber",
                "num": null
            },
            "FIGREF6": {
                "uris": null,
                "type_str": "figure",
                "text": "Proportions of three major names in individual chapters in Dream of the Red Chamber",
                "num": null
            },
            "FIGREF7": {
                "uris": null,
                "type_str": "figure",
                "text": "Bigram proportions show that Bao-Chai laughed most in early chapters",
                "num": null
            },
            "FIGREF8": {
                "uris": null,
                "type_str": "figure",
                "text": "to Historical Studies in Chinese such as DSMCTL.",
                "num": null
            },
            "TABREF0": {
                "content": "<table><tr><td>Collection</td><td>Number of Different Pseudowords</td><td>Total Number of Characters</td><td>Number of Different Characters</td><td>Number of Documents</td></tr><tr><td>Constitution</td><td>3288</td><td>713131</td><td>4097</td><td>399</td></tr><tr><td>Diplomacy</td><td>29315</td><td>2875032</td><td>5225</td><td>5758</td></tr><tr><td>Min_Bow</td><td>7784</td><td>1450623</td><td>6230</td><td>325</td></tr><tr><td>Nations</td><td>2649</td><td>679410</td><td>4916</td><td>160</td></tr><tr><td>New_People</td><td>33378</td><td>5259590</td><td>6647</td><td>1524</td></tr></table>",
                "type_str": "table",
                "text": "",
                "html": null,
                "num": null
            },
            "TABREF2": {
                "content": "<table><tr><td/><td colspan=\"2\">1) and their weights (i.e., relative importance), authors, and titles</td></tr><tr><td/><td/><td>1906</td></tr><tr><td colspan=\"3\">Weights Authors Document Title</td></tr><tr><td>420</td><td colspan=\"2\">\u6234\u9e3f\u6148 \u51fa\u4f7f\u5404\u56fd\u8003\u5bdf\u653f\u6cbb\u5927\u81e3\u6234\u9e3f\u6148\u7b49\u594f\u8bf7\u6539\u5b9a\u5168\u56fd\u5b98\u5236\u4ee5\u4e3a\uf9f7\u5baa\u9884\u5907\u6298</td></tr><tr><td>312</td><td>\u6768\u665f</td><td>\u51fa\u4f7f\u5fb7\u56fd\u5927\u81e3\u6768\u665f\u6761\u9648\u5b98\u5236\u5927\u7eb2\u6298</td></tr><tr><td>122</td><td>\u6bb7\u6d4e</td><td>\u5185\u9601\u6821\u7b7e\u4e2d\u4e66\u6bb7\u6d4e\u4e3a\u8c6b\u5907\uf9f7\u5baa\u6761\u9648\u7b79\u7ecf\u8d39\u5efa\u6d77\u519b\u7b49\u4e8c\u5341\u56db\u6761\u5448</td></tr></table>",
                "type_str": "table",
                "text": "",
                "html": null,
                "num": null
            }
        }
    }
}