| { |
| "paper_id": "2020", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T15:33:57.095193Z" |
| }, |
| "title": "A Counselling Corpus in Cantonese", |
| "authors": [ |
| { |
| "first": "John", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Hong Kong", |
| "location": {} |
| }, |
| "email": "jsylee@cityu.edu.hk" |
| }, |
| { |
| "first": "Tianyuan", |
| "middle": [], |
| "last": "Cai", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Hong Kong", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Wenxiu", |
| "middle": [], |
| "last": "Xie", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Hong Kong", |
| "location": {} |
| }, |
| "email": "wenxixie@cityu.edu.hk" |
| }, |
| { |
| "first": "Lam", |
| "middle": [], |
| "last": "Xing", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of Hong Kong", |
| "location": {} |
| }, |
| "email": "lamxing@cityu.edu.hk" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "Virtual agents are increasingly used for delivering health information in general, and mental health assistance in particular. This paper presents a corpus designed for training a virtual counsellor in Cantonese, a variety of Chinese. The corpus consists of a domainindependent subcorpus that supports small talk for rapport building with users, and a domain-specific subcorpus that provides material for a particular area of counselling. The former consists of ELIZA style responses, chitchat expressions, and a dataset of general dialog, all of which are reusable across counselling domains. The latter consists of example user inputs and appropriate chatbot replies relevant to the specific domain. In a case study, we created a chatbot with a domain-specific subcorpus that addressed 25 issues in test anxiety, with 436 inputs solicited from native speakers of Cantonese and 150 chatbot replies harvested from mental health websites. Preliminary evaluations show that Word Mover's Distance achieved 56% accuracy in identifying the issue in user input, outperforming a number of baselines.", |
| "pdf_parse": { |
| "paper_id": "2020", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "Virtual agents are increasingly used for delivering health information in general, and mental health assistance in particular. This paper presents a corpus designed for training a virtual counsellor in Cantonese, a variety of Chinese. The corpus consists of a domainindependent subcorpus that supports small talk for rapport building with users, and a domain-specific subcorpus that provides material for a particular area of counselling. The former consists of ELIZA style responses, chitchat expressions, and a dataset of general dialog, all of which are reusable across counselling domains. The latter consists of example user inputs and appropriate chatbot replies relevant to the specific domain. In a case study, we created a chatbot with a domain-specific subcorpus that addressed 25 issues in test anxiety, with 436 inputs solicited from native speakers of Cantonese and 150 chatbot replies harvested from mental health websites. Preliminary evaluations show that Word Mover's Distance achieved 56% accuracy in identifying the issue in user input, outperforming a number of baselines.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "Virtual agents are increasingly used for delivering health information in general, and mental health assistance in particular. While chatbots may not qualify to replace human counsellors, users have been found to be more willing to disclose information to a chatbot than to a human (Lucas et al., 2014) . Woebot, for example, has been shown to be effective in reducing symptoms of depression (Fitzpatrick et al., 2017) . In the \"guided conversation\" format, virtual counsellors ask pre-determined questions and allow users to choose from suggested responses (Fitzpatrick et al., 2017; Casas et al., 2018) . Other chatbots are designed to handle free-form input from users. PAL (Liu et al., 2013) and TeenChat (Huang et al., 2015) are two examples among those operating in Chinese. Targeting teenagers and young adults, these two chatbots offer advice on social topics such as family relations and love affairs. In response to user input, it selects the most relevant answer from a knowledge base, taking into account user characteristics such as gender, marital status and age. Since no single virtual counsellor can adequately address all mental health issues or serve all user populations, counselling services at schools or mental health organizations may be interested in developing their own bots to provide tailored advice in specialized counselling domains. While existing chatbots contain valuable linguistic and counselling resources, there is often no straightforward way to re-use or adapt them for related counselling tasks. To facilitate development of virtual counsellors, this paper presents a corpus that is designed to be modifiable and extendable by any counsellor, who will henceforth be referred to as the \"administrator\". Designed for chatbots in the example-based framework, it contains a domainindependent subcorpus with small talk materials for rapport building with users, an essential component in counselling sessions across all domains. It also has a domain-specific subcorpus, which is to be populated by the administrator with example user inputs and chatbot replies for the domain concerned. We developed the corpus in Cantonese, which is considered the \"most widely known and influential variety of Chinese other than Mandarin\" (Matthews and Yip, 2011) . Less frequently used in formal written communication than Mandarin, the dominant variety in mainland China, Cantonese is supported by relatively few resources for natural language processing. Although Cantonese and Mandarin are genetically related, having both developed from Middle Chinese, they are mutually unintelligible in their spoken form and have significant differences in their written form (Wong and Lee, 2018) . The differences between the major Chinese varieties have been described as being \"at least on the order of the different languages of the Romance family\" (Hannas, 1997) . The rest of the paper is organized as follows. After an overview of corpus design in the next section, we describe the content and role of the domain-independent subcorpus (Section 3) and the domain-specific subcorpus (Section 4). 1 We then present a case study on constructing a chatbot for test anxiety (Section 5).", |
| "cite_spans": [ |
| { |
| "start": 282, |
| "end": 302, |
| "text": "(Lucas et al., 2014)", |
| "ref_id": "BIBREF7" |
| }, |
| { |
| "start": 392, |
| "end": 418, |
| "text": "(Fitzpatrick et al., 2017)", |
| "ref_id": "BIBREF1" |
| }, |
| { |
| "start": 558, |
| "end": 584, |
| "text": "(Fitzpatrick et al., 2017;", |
| "ref_id": "BIBREF1" |
| }, |
| { |
| "start": 585, |
| "end": 604, |
| "text": "Casas et al., 2018)", |
| "ref_id": "BIBREF0" |
| }, |
| { |
| "start": 677, |
| "end": 695, |
| "text": "(Liu et al., 2013)", |
| "ref_id": "BIBREF6" |
| }, |
| { |
| "start": 709, |
| "end": 729, |
| "text": "(Huang et al., 2015)", |
| "ref_id": "BIBREF3" |
| }, |
| { |
| "start": 2260, |
| "end": 2284, |
| "text": "(Matthews and Yip, 2011)", |
| "ref_id": "BIBREF8" |
| }, |
| { |
| "start": 2688, |
| "end": 2708, |
| "text": "(Wong and Lee, 2018)", |
| "ref_id": "BIBREF13" |
| }, |
| { |
| "start": 2865, |
| "end": 2879, |
| "text": "(Hannas, 1997)", |
| "ref_id": "BIBREF2" |
| }, |
| { |
| "start": 3113, |
| "end": 3114, |
| "text": "1", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1." |
| }, |
| { |
| "text": "Our corpus, which consists of post-reply sentence pairs, is designed to support the construction of virtual counsellors in any domain. Given a user input, a chatbot can identify the example post in the corpus that is semantically most similar, and then retrieve the corresponding reply. According to the five-stage framework for counselling sessions adopted by Inoue et al. (2012) , the Introduction stage establishes rapport between the therapist and the client. In the Elaboration stage, the therapist explores the client's situation and tries to find clues for a solution. This may be followed by a Resistance stage if the client feels uncomfortable or resists the therapy. The therapist then proposes actions toward a solution in the Intervention stage. Finally, the Solution stage concludes the session. Table 2 : Size of the domain-independent subcorpus (Section 3), with breakdown into the main categories", |
| "cite_spans": [ |
| { |
| "start": 361, |
| "end": 380, |
| "text": "Inoue et al. (2012)", |
| "ref_id": "BIBREF4" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 809, |
| "end": 816, |
| "text": "Table 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Corpus Design", |
| "sec_num": "2." |
| }, |
| { |
| "text": "The domain-independent subcorpus (Section 3) provides content for \"off-task,\" social exchanges that are common across many counselling domains, especially during the Introduction stage. In subsequent stages, as users discuss their specific issues, the chatbot is expected to detect the issue types and incorporate appropriate advice in its response. For this purpose, the domain-specific subcorpus (Section 4) is to be populated by the administrator with example user inputs and possible advice to address them.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Category", |
| "sec_num": null |
| }, |
| { |
| "text": "The domain-independent subcorpus targets off-task interactions, such as greetings and chitchat. These interactions are typically dominant in the Introduction stage of a counselling session, but they can also be interweaved with ontask interactions in the subsequent stages. Their purpose is to move the conversation along, while keeping the users engaged and preparing them to disclose their feelings and issues. Several example post-reply pairs in this subcorpus are shown in Table 1 . Expected to be re-usable for most counselling domains and target users, these pairs belong to one of the four categories below. Table 2 shows the size of the domain-independent subcorpus with a breakdown into these categories.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 477, |
| "end": 484, |
| "text": "Table 1", |
| "ref_id": "TABREF1" |
| }, |
| { |
| "start": 615, |
| "end": 622, |
| "text": "Table 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Domain-independent Subcorpus", |
| "sec_num": "3." |
| }, |
| { |
| "text": "The chitchat category covers common user expressions such as greetings, thank-you and good-bye, to which formulaic responses from the chatbot is usually sufficient; short yes/no answers (e.g., \u4fc2 \u5440 'yes') to the preceding question; as well as small talk such as user enquiries about the chatbot's personal information, including name, age, occupation, family members, favorite food or colors, etc. The chatbot's persona is the only content that is expected to be customized in this subcorpus.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Chitchat", |
| "sec_num": "3.1." |
| }, |
| { |
| "text": "Encouragers consist of backchannel or empathic replies, mostly defined with ELIZA-like regular expressions (Weizenbaum, 1983) .", |
| "cite_spans": [ |
| { |
| "start": 107, |
| "end": 125, |
| "text": "(Weizenbaum, 1983)", |
| "ref_id": "BIBREF12" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Encouragers", |
| "sec_num": "3.2." |
| }, |
| { |
| "text": "Advice requests are user inputs that explicitly ask the chatbot for counselling advice.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Advice Requests", |
| "sec_num": "3.3." |
| }, |
| { |
| "text": "If the user input does not match any of the above categories, the chatbot can backoff to a large-scale dialog database.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Dialog", |
| "sec_num": "3.4." |
| }, |
| { |
| "text": "We compiled a set of over 276,000 Cantonese and Chinese post-reply pairs, to handle general or off-task user input. These pairs are taken from two conversation corpora: the Xiaohuangji corpus 2 and the ChatterBot corpus 3 .", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Dialog", |
| "sec_num": "3.4." |
| }, |
| { |
| "text": "Following the Introduction stage of the counselling session, the chatbot engages users in discussing their thoughts and experiences, and then gives counselling advice. The domain-specific subcorpus consists of example user inputs and appropriate pieces of advice. Unlike those in the domain-independent subcorpus (Section 3), these postreply pairs are less likely to be relevant to other domains. The onus is therefore put on the administrators to supply these pairs for their target counselling domain. Most existing virtual counsellors clearly define their areas of competence. For example, PAL addresses the topics \"husband and wife\", \"family relations\", \"love affairs\", \"adolescence\", \"feeling and mood\", and \"mental tutors\" (Liu et al., 2013) , while TeenChat detect stress in the areas of \"study\", \"self-cognition\", \"interpersonal\", \"affection\" and \"general\" (Huang et al., 2015) . The domainspecific subcorpus likewise expects the administrator to define a set of issues, but at a finer granularity, such that each issue can be mapped to specific pieces of advice.", |
| "cite_spans": [ |
| { |
| "start": 729, |
| "end": 747, |
| "text": "(Liu et al., 2013)", |
| "ref_id": "BIBREF6" |
| }, |
| { |
| "start": 865, |
| "end": 885, |
| "text": "(Huang et al., 2015)", |
| "ref_id": "BIBREF3" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Domain-specific Subcorpus", |
| "sec_num": "4." |
| }, |
| { |
| "text": "We will refer to each issue as a \"symptom\". In the domain of test anxiety, for example, symptoms may include headaches, worries about failure, and worries about parental reaction, which are among the most relevant issues identified in a psychology study (Wren and Benson, 2004) . For each symptom, the administrator is to provide a set of symptom statements, i.e., typical user inputs that express that symptom. Table 3 shows example symptom statements for the domain of test anxiety (Wren and Benson, 2004) .", |
| "cite_spans": [ |
| { |
| "start": 254, |
| "end": 277, |
| "text": "(Wren and Benson, 2004)", |
| "ref_id": "BIBREF14" |
| }, |
| { |
| "start": 484, |
| "end": 507, |
| "text": "(Wren and Benson, 2004)", |
| "ref_id": "BIBREF14" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 412, |
| "end": 419, |
| "text": "Table 3", |
| "ref_id": "TABREF3" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Symptom Statements", |
| "sec_num": "4.1." |
| }, |
| { |
| "text": "The chatbot should be able to address each symptom with appropriate pieces of advice, which we will refer to as \"counselling items\". The administrator is to provide these items, which may be practical advice collected from existing mental health materials. They can also be contact information of human counsellors, which would be appropriate for more severe symptoms such as thoughts of suicide. The administrator can optionally provide a set of sentences for broaching the counselling topic 4 , to be deployed when users engage exclusively in small talk for an extended period, to steer them back to on-task discussion.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Counselling Items", |
| "sec_num": "4.2." |
| }, |
| { |
| "text": "The administrator may adjust the aggressiveness of the chatbot in advice giving by setting the threshold score for symptom detection. The typical system computes a similarity score between the user input and each symptom statement in the subcorpus (Section 5.2). If the score exceeds the threshold, a counselling item for the symptom is returned. A higher threshold leads to a chatbot that is more restrained in giving advice and more inclined to listen. A lower threshold, in contrast, increases its propensity to give counsel.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Threshold Score", |
| "sec_num": "4.3." |
| }, |
| { |
| "text": "To evaluate the proposed chatbot framework, we compiled a domain-specific subcorpus for the domain of test anxiety (Section 5.1) and experimented with a number of semantic similarity measures (Section 5.2). Together with the domain-independent subcorpus, we instantiated a chatbot that addresses test anxiety (Figure 1 ). Many websites and pamphlets already offer advice and remedies for test anxiety, an issue that affects many students. Despite the wealth of counselling materials static media, however, there has been little attempt to leverage them for virtual counselling. To the best of our knowledge, the only reported chatbot that specifically deals with test 4 e.g., \u4f60\u6709\u5187\u95dc\u65bc\u5b78\u696d\u65e2\u91ce\u60f3\u540c\u6211\u50be\u4e0b\uff1f 'Do you want to talk about any academic issue?' Figure 1 : Example conversation in our Cantonese chatbot for test anxiety anxiety is the Exam-Stress Counselor and Academic Planner (Rudra et al., 2012) . This system uses a deterministic finite state automaton to analyze user inputs, which are expected to conform to a number of fixed patterns. Our chatbot, in contrast, is designed for interactions with free-form text.", |
| "cite_spans": [ |
| { |
| "start": 872, |
| "end": 892, |
| "text": "(Rudra et al., 2012)", |
| "ref_id": "BIBREF11" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 309, |
| "end": 318, |
| "text": "(Figure 1", |
| "ref_id": null |
| }, |
| { |
| "start": 740, |
| "end": 748, |
| "text": "Figure 1", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Case Study", |
| "sec_num": "5." |
| }, |
| { |
| "text": "Our set of symptoms were the 25 most relevant issues in test anxiety identified by Wren and Benson (2004) . For symptom statements, we solicited user inputs for each symptom from 12 subjects, all native speakers of Cantonese. After reading the original symptom description in English written by Wren and Benson (2004) , the subjects were asked to give paraphrases in Cantonese. They provided a total of 436 sentences, which served as the evaluation dataset. For counselling items, we collected advice on test anxiety from websites linked to the home page of the counselling department at our university. For severe symptoms requiring human intervention, we supplied the phone number of the department. The subcorpus contained a total of 150 counselling items, with an average of 5.8 items per symptom.", |
| "cite_spans": [ |
| { |
| "start": 83, |
| "end": 105, |
| "text": "Wren and Benson (2004)", |
| "ref_id": "BIBREF14" |
| }, |
| { |
| "start": 295, |
| "end": 317, |
| "text": "Wren and Benson (2004)", |
| "ref_id": "BIBREF14" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Compilation of Domain-specific Subcorpus", |
| "sec_num": "5.1." |
| }, |
| { |
| "text": "To determine the symptom in the user input, the chatbot computes semantic similarity scores between the input and", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semantic Similarity Measures", |
| "sec_num": "5.2." |
| }, |
| { |
| "text": "Accuracy Word Mover's Distance 56.2% ELMo 47.5% word2vec 47.7% Table 4 : Symptom detection accuracy the symptom statements in the domain-specific subcorpus. We compared the performance of two semantic similarity measures with the baseline of using word2vec (Mikolov et al., 2013) embeddings alone. Word Mover's Distance (WMD). We applied this distance metric by representing each word in the sentence with the word2vec vector embeddings. We then used WMD to measure the dissimilarity between two sentences, as expressed by the minimum cumulative distance that the embedded words of one sentence need to travel to match those in the other sentence (Kusner et al., 2015) . The symptom statement that yields the shortest distance is returned.", |
| "cite_spans": [ |
| { |
| "start": 257, |
| "end": 279, |
| "text": "(Mikolov et al., 2013)", |
| "ref_id": "BIBREF9" |
| }, |
| { |
| "start": 647, |
| "end": 668, |
| "text": "(Kusner et al., 2015)", |
| "ref_id": "BIBREF5" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 63, |
| "end": 70, |
| "text": "Table 4", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Approach", |
| "sec_num": null |
| }, |
| { |
| "text": "ELMo. Shown to improve the state-of-the-art in a variety of NLP tasks, ELMo is a contextualized word representation that models not only the characteristics of word usage, but also how the usage varies across linguistic contexts, i.e., polysemy (Peters et al., 2018) . The word vectors are learned functions of the internal states of a deep bidirectional language model, pre-trained on a large text corpus.", |
| "cite_spans": [ |
| { |
| "start": 245, |
| "end": 266, |
| "text": "(Peters et al., 2018)", |
| "ref_id": "BIBREF10" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Approach", |
| "sec_num": null |
| }, |
| { |
| "text": "We conducted a leave-one-out evaluation on the symptom statements in our domain-specific subcorpus (Section 5.1) to measure accuracy in symptom detection. The chatbot automatically determined the symptom for each of the 436 statements, by retrieving the most similar statement among those in the remainder of the subcorpus, with respect to the similarity measures described in Section 5.2. As shown in Table 4 , Word Mover's Distance achieved the best performance in symptom detection, at 56.2% accuracy. It outperformed both ELMo (47.5%) and the word2vec baseline (47.7%).", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 402, |
| "end": 409, |
| "text": "Table 4", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Symptom Detection Accuracy", |
| "sec_num": "5.3." |
| }, |
| { |
| "text": "We have presented the first text corpus that is designed for virtual counselling in Cantonese. The corpus consists of a domain-independent subcorpus of chitchat and general dialog materials, which are intended to be re-usable for any counselling domain; and a domain-specific subcorpus, to be populated by administrators with typical user inputs and counselling advice in the domain concerned. We reported a case study on test anxiety, and evaluated the chatbot's accuracy in symptom detection. It is hoped that this corpus will facilitate development of virtual counsellors to serve Cantonese speakers.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusions", |
| "sec_num": "6." |
| }, |
| { |
| "text": "The corpus is available for research purposes on request to the first author.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Accessed at https://github.com/skdjfla/dgk lost conv 3 Accessed at https://github.com/gunthercox/chatterbot-corpus", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "This work was partially supported by CityU Internal Funds for ITF Projects (no. 9678104).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Acknowledgements", |
| "sec_num": "7." |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Food Diary Coaching Chatbot", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Casas", |
| "suffix": "" |
| }, |
| { |
| "first": "E", |
| "middle": [], |
| "last": "Mugellini", |
| "suffix": "" |
| }, |
| { |
| "first": "O", |
| "middle": [ |
| "A" |
| ], |
| "last": "Khaled", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proc. ACM International Joint Conference and International Symposium on Pervasive and Ubiquitous Computing and Wearable Computers (UbiComp)", |
| "volume": "", |
| "issue": "", |
| "pages": "1676--1680", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Casas, J., Mugellini, E., and Khaled, O. A. (2018). Food Diary Coaching Chatbot. In Proc. ACM International Joint Conference and International Symposium on Per- vasive and Ubiquitous Computing and Wearable Com- puters (UbiComp), pages 1676-1680.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Delivering Cognitive Behavior Therapy to Young Adults with Symptoms of Depression and Anxiety using a Fully Automated Conversational Agent (Woebot): A Randomized Controlled Trial", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [ |
| "K" |
| ], |
| "last": "Fitzpatrick", |
| "suffix": "" |
| }, |
| { |
| "first": "A", |
| "middle": [], |
| "last": "Darcy", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Vierhile", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "JMIR Mental Health", |
| "volume": "4", |
| "issue": "2", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Fitzpatrick, K. K., Darcy, A., and Vierhile, M. (2017). De- livering Cognitive Behavior Therapy to Young Adults with Symptoms of Depression and Anxiety using a Fully Automated Conversational Agent (Woebot): A Random- ized Controlled Trial. JMIR Mental Health, 4(2):e19.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Asia's Orthographic Dilemma", |
| "authors": [ |
| { |
| "first": "W", |
| "middle": [ |
| "C" |
| ], |
| "last": "Hannas", |
| "suffix": "" |
| } |
| ], |
| "year": 1997, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Hannas, W. C. (1997). Asia's Orthographic Dilemma. University of Hawaii Press, Honolulu, HI.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "TeenChat: A Chatterbot System for Sensing and Releasing Adolescents' Stress", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Huang", |
| "suffix": "" |
| }, |
| { |
| "first": "Q", |
| "middle": [], |
| "last": "Li", |
| "suffix": "" |
| }, |
| { |
| "first": "Y", |
| "middle": [], |
| "last": "Xue", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Cheng", |
| "suffix": "" |
| }, |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Xu", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Jia", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [], |
| "last": "Feng", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "", |
| "volume": "9085", |
| "issue": "", |
| "pages": "133--145", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Huang, J., Li, Q., Xue, Y., Cheng, T., Xu, S., Jia, J., and Feng, L. (2015). TeenChat: A Chatterbot System for Sensing and Releasing Adolescents' Stress. LNCS, 9085:133-145.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Multimodal Corpus for Psychotherapeutic Situation", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Inoue", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Hanada", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Furuyama", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Irino", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Ichinomiya", |
| "suffix": "" |
| }, |
| { |
| "first": "H", |
| "middle": [], |
| "last": "Massaki", |
| "suffix": "" |
| } |
| ], |
| "year": 2012, |
| "venue": "Proc. LREC Workshop on Multimodal Corpora for Machine Learning", |
| "volume": "", |
| "issue": "", |
| "pages": "18--21", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Inoue, M., Hanada, R., Furuyama, N., Irino, T., Ichinomiya, T., and Massaki, H. (2012). Multimodal Corpus for Psy- chotherapeutic Situation. In Proc. LREC Workshop on Multimodal Corpora for Machine Learning, pages 18- 21.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "From word embeddings to document distances", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Kusner", |
| "suffix": "" |
| }, |
| { |
| "first": "Y", |
| "middle": [], |
| "last": "Sun", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Kolkin", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Weinberger", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "Proc. International Conference on Machine Learning", |
| "volume": "", |
| "issue": "", |
| "pages": "957--966", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kusner, M., Sun, Y., Kolkin, N., and Weinberger, K. (2015). From word embeddings to document distances. In Proc. International Conference on Machine Learning, pages 957-966.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "PAL: A Chatterbot System for Answering Domainspecific Questions", |
| "authors": [ |
| { |
| "first": "Y", |
| "middle": [], |
| "last": "Liu", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Liu", |
| "suffix": "" |
| }, |
| { |
| "first": "X", |
| "middle": [], |
| "last": "Wang", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [], |
| "last": "Wang", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Li", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Proc. 51st Annual Meeting of the Association for Computational Linguistics (ACL)", |
| "volume": "", |
| "issue": "", |
| "pages": "67--72", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Liu, Y., Liu, M., Wang, X., Wang, L., and Li, J. (2013). PAL: A Chatterbot System for Answering Domain- specific Questions. In Proc. 51st Annual Meeting of the Association for Computational Linguistics (ACL), pages 67-72.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "It's Only a Computer: Virtual Humans Increase Willingness to Disclose", |
| "authors": [ |
| { |
| "first": "G", |
| "middle": [ |
| "M" |
| ], |
| "last": "Lucas", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Gratch", |
| "suffix": "" |
| }, |
| { |
| "first": "A", |
| "middle": [], |
| "last": "King", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [ |
| "P" |
| ], |
| "last": "Morency", |
| "suffix": "" |
| } |
| ], |
| "year": 2014, |
| "venue": "Computers in Human Behavior", |
| "volume": "37", |
| "issue": "", |
| "pages": "94--100", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lucas, G. M., Gratch, J., King, A., and Morency, L. P. (2014). It's Only a Computer: Virtual Humans Increase Willingness to Disclose. Computers in Human Behavior, 37:94-100.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Cantonese: A Comprehensive Grammar", |
| "authors": [ |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Matthews", |
| "suffix": "" |
| }, |
| { |
| "first": "V", |
| "middle": [], |
| "last": "Yip", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Matthews, S. and Yip, V. (2011). Cantonese: A Compre- hensive Grammar. Routledge, New York.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "Efficient Estimation of Word Representations in Vector Space", |
| "authors": [ |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Mikolov", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "" |
| }, |
| { |
| "first": "G", |
| "middle": [], |
| "last": "Corrado", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Dean", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Proc. International Conference on Learning Representations", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mikolov, T., Chen, K., Corrado, G., and Dean, J. (2013). Efficient Estimation of Word Representations in Vector Space. In Proc. International Conference on Learning Representations (ICLR).", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "Deep Contextualized Word Representations", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [ |
| "E" |
| ], |
| "last": "Peters", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Neumann", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Iyyer", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Gardner", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Clark", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [], |
| "last": "Zettlemoyer", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proc. NAACL-HLT", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Peters, M. E., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., and Zettlemoyer, L. (2018). Deep Contex- tualized Word Representations. In Proc. NAACL-HLT.", |
| "links": null |
| }, |
| "BIBREF11": { |
| "ref_id": "b11", |
| "title": "ESCAP: Towards the Design of an AI Architecture for a Virtual Counselor to Tackle Students' Exam Stress", |
| "authors": [ |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Rudra", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Li", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Kavakli", |
| "suffix": "" |
| } |
| ], |
| "year": 2012, |
| "venue": "Proc. 45th Hawaii International Conference on System Sciences", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Rudra, T., Li, M., and Kavakli, M. (2012). ESCAP: To- wards the Design of an AI Architecture for a Virtual Counselor to Tackle Students' Exam Stress. In Proc. 45th Hawaii International Conference on System Sci- ences.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "ELIZA-A Computer Program For the Study of Natural Language Communication Between Man and Machine", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Weizenbaum", |
| "suffix": "" |
| } |
| ], |
| "year": 1983, |
| "venue": "Communications of the ACM", |
| "volume": "26", |
| "issue": "1", |
| "pages": "23--28", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Weizenbaum, J. (1983). ELIZA-A Computer Program For the Study of Natural Language Communication Be- tween Man and Machine. Communications of the ACM, 26(1):23-28.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "Register-sensitive Translation: A Case Study of Mandarin and Cantonese", |
| "authors": [ |
| { |
| "first": "T.-S", |
| "middle": [], |
| "last": "Wong", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proc. Association for Machine Translation in the Americas (AMTA)", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Wong, T.-S. and Lee, J. (2018). Register-sensitive Trans- lation: A Case Study of Mandarin and Cantonese. In Proc. Association for Machine Translation in the Ameri- cas (AMTA).", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "Measuring Test Anxiety in Children: Scale Development and Internal Construct Validation", |
| "authors": [ |
| { |
| "first": "D", |
| "middle": [ |
| "G" |
| ], |
| "last": "Wren", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Benson", |
| "suffix": "" |
| } |
| ], |
| "year": 2004, |
| "venue": "Stress, & Coping: An international Journal", |
| "volume": "17", |
| "issue": "3", |
| "pages": "227--240", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Wren, D. G. and Benson, J. (2004). Measuring Test Anx- iety in Children: Scale Development and Internal Con- struct Validation. Anxiety, Stress, & Coping: An interna- tional Journal, 17(3):227-240.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "TABREF0": { |
| "num": null, |
| "type_str": "table", |
| "html": null, |
| "content": "<table><tr><td/><td>Description</td><td>Example user input</td><td>Example chatbot reply</td></tr><tr><td>Chitchat</td><td>Short social</td><td>\u4f60\u597d 'hi'</td><td>\u4f60\u597d 'hi'</td></tr><tr><td/><td>expressions</td><td>\u5514\u8a72 'thanks'</td><td/></tr><tr><td>Advice</td><td colspan=\"2\">Explicit request for \u6211\u60f3\u807d\u4e0b\u4f60\u65e2\u5efa\u8b70 'I want to hear</td><td>(To be drawn from the</td></tr><tr><td>requests</td><td>advice</td><td>your advice'</td><td>domain-specific subcorpus)</td></tr></table>", |
| "text": "\u5514\u6d17\u5ba2\u6c23 'you're welcome' \u62dc\u62dc 'goodbye' \u5f97\u9592\u518d\u6435\u6211\u5566 'see you later' Small talk \u4f60\u53eb\u54a9\u540d\uff1f 'What's your name?' \u6211\u53eb\u505a... 'My name is ...' Dialog General dialog \u4f60\u7684\u5de5\u4f5c\u662f\u4ec0\u9ebc\uff1f 'What's your job?' \u966a\u804a 'Talking to people.' Encouragers ELIZA expressions \u6211\u89ba\u5f97 ... 'I feel ...' \u9ede\u89e3\u89ba\u5f97 ...? 'Why do you feel ... ?'" |
| }, |
| "TABREF1": { |
| "num": null, |
| "type_str": "table", |
| "html": null, |
| "content": "<table><tr><td/><td>: Domain-independent subcorpus (Section 3): example post-reply pairs</td></tr><tr><td>Category</td><td># post-reply pairs</td></tr><tr><td>Chitchat</td><td>123</td></tr><tr><td>Advice requests</td><td>41</td></tr><tr><td>Encouragers</td><td>103</td></tr><tr><td>Dialog</td><td>276,000</td></tr></table>", |
| "text": "" |
| }, |
| "TABREF3": { |
| "num": null, |
| "type_str": "table", |
| "html": null, |
| "content": "<table/>", |
| "text": "Domain-specific corpus (Section 4): example symptoms, symptom statements and counselling items for the test anxiety domain" |
| } |
| } |
| } |
| } |
