| { |
| "paper_id": "S16-1025", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T15:26:45.195967Z" |
| }, |
| "title": "NRU-HSE at SemEval-2016 Task 4: Comparative Analysis of Two Iterative Methods Using Quantification Library", |
| "authors": [ |
| { |
| "first": "Nikolay", |
| "middle": [], |
| "last": "Karpov", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "nkarpov@hse.ru" |
| }, |
| { |
| "first": "Alexander", |
| "middle": [], |
| "last": "Porshnev", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "aporshnev@hse.ru" |
| }, |
| { |
| "first": "Kirill", |
| "middle": [], |
| "last": "Rudakov", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "rudakovkirillx@gmail.com" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "In many areas, such as social science, politics or market research, people need to track sentiment and their changes over time. For sentiment analysis in this field it is more important to correctly estimate proportions of each sentiment expressed in the set of documents (quantification task) than to accurately estimate sentiment of a particular document (classification). Basically, our study was aimed to analyze the effectiveness of two iterative quantification techniques and to compare their effectiveness with baseline methods. All the techniques are evaluated using a set of synthesized data and the SemEval-2016 Task4 dataset. We made the quantification methods from this paper available as a Python open source library. The results of comparison and possible limitations of the quantification techniques are discussed.", |
| "pdf_parse": { |
| "paper_id": "S16-1025", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "In many areas, such as social science, politics or market research, people need to track sentiment and their changes over time. For sentiment analysis in this field it is more important to correctly estimate proportions of each sentiment expressed in the set of documents (quantification task) than to accurately estimate sentiment of a particular document (classification). Basically, our study was aimed to analyze the effectiveness of two iterative quantification techniques and to compare their effectiveness with baseline methods. All the techniques are evaluated using a set of synthesized data and the SemEval-2016 Task4 dataset. We made the quantification methods from this paper available as a Python open source library. The results of comparison and possible limitations of the quantification techniques are discussed.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "In many areas, such as customer-relationship management or opinion mining, people need to track changes over time and measure proportions of documents expressing different sentiments. In these situations, the task of accurate categorization of each document is replaced by the task of providing accurate proportions of documents from each class (quantification). George Forman suggested defining the 'quantification task' as finding the best estimate for the amount of cases in each class in a test set, using a training set with substantially different class distribution (Forman, 2008) .", |
| "cite_spans": [ |
| { |
| "start": 573, |
| "end": 587, |
| "text": "(Forman, 2008)", |
| "ref_id": "BIBREF6" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "Application of the quantification approach in opinion mining , network-behavior analysis (Tang et al., 2010) , word-sense disambiguation (Chan and Ng, 2006) , remote sensing (Guerrero-Curieses et al., 2009) , quality control (S\u00e1nchez et al., 2008) , monitoring support-call logs (Forman et al., 2006) and credit scoring (Hand and others, 2006) showed high performance even with a relatively small training set.", |
| "cite_spans": [ |
| { |
| "start": 89, |
| "end": 108, |
| "text": "(Tang et al., 2010)", |
| "ref_id": "BIBREF15" |
| }, |
| { |
| "start": 137, |
| "end": 156, |
| "text": "(Chan and Ng, 2006)", |
| "ref_id": "BIBREF2" |
| }, |
| { |
| "start": 174, |
| "end": 206, |
| "text": "(Guerrero-Curieses et al., 2009)", |
| "ref_id": "BIBREF9" |
| }, |
| { |
| "start": 225, |
| "end": 247, |
| "text": "(S\u00e1nchez et al., 2008)", |
| "ref_id": "BIBREF14" |
| }, |
| { |
| "start": 279, |
| "end": 300, |
| "text": "(Forman et al., 2006)", |
| "ref_id": "BIBREF7" |
| }, |
| { |
| "start": 320, |
| "end": 343, |
| "text": "(Hand and others, 2006)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "Although quantification techniques are able to provide accurate sentiment analysis of proportions in situations of distribution drift, the question of optimal technique for analysis of tweets still raises a lot of questions. It is worth mentioning that sentiment analysis of tweets presents additional challenges to natural language processing, because of the small amount of text (less than 140 characters in each document), usage of creative spelling (e.g. \"happpyyy\", \"some1 yg bner2 tulus\"), abbreviations (such as \"wth\" or \"lol\"), informal constructions (\"hahahaha yava quiet so !ma I m bored av even home nw\") and hashtags (BREAKING: US GDP growth is back! #kidding), which are a type of tagging for Twitter messages.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "In our paper we used several quantification methods mentioned in literature as the best ones and evaluated them by comparing their effectiveness with one another and with baseline methods.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "The paper is organized as follows. In Section 2, we first look at the notation, then we briefly overview six methods to solve the quantification problem. Section 3 describes two datasets we use in our research. Section 4 describes the results of our experiments, while Section 5 concludes the work defining open research issues for further investigation.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "In this section we describe the methods used to handle changes in class distribution.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": "First, let us give some definition of notation. \u0425: vector representation of observation x; C = {c 1 , \u2026, c n }: classes of observations, where n is the number of classes;", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": "(c): a true prior probability (aka \"prevalence\" of class c in the set S; \u0302 (c j ): estimated prevalence of c j using the set S; \u0302 (c j ): estimated \u0302 (c j ) obtained via method M; p(c j /x): a posteriori probabilitiesto classify an observation x to the class c j ;", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": ",", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": ": training and test sets of observations, respectively;", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": ": a subset of set where each observation falls within class ; _ = {p TEST (c i )}; i=1, : class probability distribution of the test set; _ = {p TRAIN (c i )}; i=1, : class probability distribution of the training set;", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": "The problem we study has some training set, which provides us with a set of labeled examples -TRAIN, with class distribution TRAIN_CD. At some point the distribution of data changes to a new, but unknown class distribution -TEST_CD, and this distribution provides a set of unlabeled examples -TEST. Given this terminology, we can state our quantification problem more precisely.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Quantification Methods", |
| "sec_num": "2" |
| }, |
| { |
| "text": "The first approach provides information about proportions of document in each class just by classification of each document. In this case, the process starts with training the best available classifier, applying it to the test set and counting the amount of documents in each class. Forman named this obvious approach as Classify and Count (CC) (Forman, 2008) .", |
| "cite_spans": [ |
| { |
| "start": 345, |
| "end": 359, |
| "text": "(Forman, 2008)", |
| "ref_id": "BIBREF6" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Classify and Count", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "The observed count P of positives from the classifier will include both true positives and false positives, P = TP + FP, as characterized by the standard 2 \u00d7 2 confusion matrix.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Classify and Count", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "Classifier Predictions:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Classify and Count", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "Actual\\Prediction P_ N_ P TP FN N FP TN", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Classify and Count", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "Adjusted Classify and Count (ACC -aka the \"confusion matrix model\" quantification method (Forman, 2005) consists of six steps: 1. training a binary classifier on the entire training set 2. estimating its characteristics via many-fold cross-validation (tpr = TP/P and fpr = FP/N) 3. applying the classifier to the test set 4. counting the number of test cases on which the classifier outputs positives 5. estimating the true percentage of positives via Equation (1)", |
| "cite_spans": [ |
| { |
| "start": 89, |
| "end": 103, |
| "text": "(Forman, 2005)", |
| "ref_id": "BIBREF5" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Adjusted Classify and Count", |
| "sec_num": "2.2" |
| }, |
| { |
| "text": "EQUATION", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [ |
| { |
| "start": 0, |
| "end": 8, |
| "text": "EQUATION", |
| "ref_id": "EQREF", |
| "raw_str": "\u0302 ( ) = (", |
| "eq_num": ") ( ) ( ) ( ) (1)" |
| } |
| ], |
| "section": "Adjusted Classify and Count", |
| "sec_num": "2.2" |
| }, |
| { |
| "text": "6. clipping the output to the feasible range.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Adjusted Classify and Count", |
| "sec_num": "2.2" |
| }, |
| { |
| "text": "As mentioned by Forman, the performance of the ACC method degrades severely in the situation of a highly imbalanced training sample. If one of the classes is rare in the training set, the classifier will learn not to vote for this class because of tpr = 0%. Small denominator (tpr \u2212 fpr) in Equation 1makes the quotient highly sensitive in the estimation of tpr or fpr, and this leads to low quantification accuracy especially at the small training sets with high class imbalance (Forman et al., 2006) .", |
| "cite_spans": [ |
| { |
| "start": 480, |
| "end": 501, |
| "text": "(Forman et al., 2006)", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Adjusted Classify and Count", |
| "sec_num": "2.2" |
| }, |
| { |
| "text": "The Probabilistic Classify and Count (PCC) method differs from the CC algorithm by counting the expected share of positive predicted documents, i.e. the probability of membership in class c of observation after classifying documents in the TEST set.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Probabilistic Classify and Count", |
| "sec_num": "2.3" |
| }, |
| { |
| "text": "\u0302 ( ) = \u2211 ( | ) \u2208 | |", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Probabilistic Classify and Count", |
| "sec_num": "2.3" |
| }, |
| { |
| "text": "(2)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Probabilistic Classify and Count", |
| "sec_num": "2.3" |
| }, |
| { |
| "text": "The central idea of the Probabilistic Adjusted Classify and Count (PACC) algorithm is evidently to combine two algorithms above -ACC and PCC. \u0302 ( ), ( ), ( ) should be replaced by their expected values, i.e.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Probabilistic Adjusted Classify and Count", |
| "sec_num": "2.4" |
| }, |
| { |
| "text": "EQUATION", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [ |
| { |
| "start": 0, |
| "end": 8, |
| "text": "EQUATION", |
| "ref_id": "EQREF", |
| "raw_str": "\u0302 ( )~ \u0302 ( ), ( )~ { ( )}, ( )~ { ( )}, where { ( )} = \u2211 ( | ) \u2208 | | { ( )} = \u2211 ( | ) \u2208 | \u0305 | then the form of the PACC is \u0302 ( ) = ( ) { ( )} { ( )} { ( )}", |
| "eq_num": "(3)" |
| } |
| ], |
| "section": "Probabilistic Adjusted Classify and Count", |
| "sec_num": "2.4" |
| }, |
| { |
| "text": "A simple procedure to adjust the outputs of a classifier to a new a priori probability is described in the study by (Saerens et al., 2002) .", |
| "cite_spans": [ |
| { |
| "start": 116, |
| "end": 138, |
| "text": "(Saerens et al., 2002)", |
| "ref_id": "BIBREF13" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "EQUATION", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [ |
| { |
| "start": 0, |
| "end": 8, |
| "text": "EQUATION", |
| "ref_id": "EQREF", |
| "raw_str": "( / ) = ( / ) \u2211 ( / )", |
| "eq_num": "(4)" |
| } |
| ], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "It is important that authors suggest using not only the well-known formula (4) to compute the corrected a posteriori probabilities, but also an iterative procedure to adjust the outputs of the trained classier with respect to these new a priori probabilities, without having to refit the model, even when these probabilities are not known in advance.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "To make the Expectation Maximization (EM) method clear, we specify its algorithm in Figure1 using a pseudo-code. The algorithm begins with counting start values for class probability distribution, using labels on the training set TRAIN (line 1), builds an initial classifier C_i from the TRAIN set (line 2) and classifies each item in the unlabeled TEST set (line3), where the classify functions return the a posteriori probabilities (TEST_prob) for the specified datasets. The algorithm then iterates in lines 4-9 until the maximum number of iterations (maxIterations) is reached. In this loop, the algorithm first uses the previous a posteriori probabilities TEST_prob to estimate a new a priori probability (line 6). Then, in line 7, a posteriori probabilities are computed using Equation (4). Finally, once the loop terminates, the last posteriori probabilities returns (line 9).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "EM (TRAIN, TEST) 1.TEST_CD = prevalence(TRAIN) 2. C_i = build_clf(TRAIN) 3. TEST_prob = classify(C_i, TEST) 4. for (i=1; i<maxIterations; i++) 5. { 6. TEST_CD = prevalence(TEST_prob) 7. TEST_prob = bayes(TEST_CD, TEST_prob) 8. } 9. return TEST_CD Figure 1 : Pseudo-code for the EM algorithm.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 247, |
| "end": 255, |
| "text": "Figure 1", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "To build a classifier in the function build_clf, we use support vector machines (SVM) with linear kernel.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Expectation Maximization", |
| "sec_num": "2.5" |
| }, |
| { |
| "text": "Another interesting method is iterative cost-sensitive class distribution estimation (CDEIterate) described in the study by (Xue and Weiss, 2009) .", |
| "cite_spans": [ |
| { |
| "start": 124, |
| "end": 145, |
| "text": "(Xue and Weiss, 2009)", |
| "ref_id": "BIBREF16" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "The main idea of this method is to retrain a classifier at each iteration, where the iterations progressively improve the quantification accuracy of performing the \u00abclassify and count\u00bb method via the generated costsensitive classifiers.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "For the CDE-based method, the final prevalence is induced from the TRAIN labeled set with the cost of classes COST. The COST value is computed with Equation (5), utilizing the class distribution calculated during the previous step TEST_CD. For each iteration, we recalculate:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "= _ _ (5)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "The CDEIterate algorithm is specified in Figure 2 , using the pseudo-code. The algorithm begins with counting the class distribution TRAIN_CD for training labels TRAIN (line 1). Then it builds an initial classifier C_i from the TRAIN set (line 2). In a loop, this algorithm uses the previous classifier C_i to classify the unlabeled TEST set by estimating a posterior probability TEST_prob for each item in a test set (line 5). Then. in line 6, the a priory probability distribution is computed and the cost ratio information is updated (line 7). In line 8, a new cost-sensitive classifier C_i is generated using the TRAIN set with the updated cost ratioCOST. The algorithm then iterates in lines 4-9 until the maximum number of iterations (maxIterations) is reached. Finally, once the loop terminates, the last a priory probability distribution of classes is returned TEST_CD (line 10).", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 41, |
| "end": 49, |
| "text": "Figure 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "CDEIterate (TRAIN, TEST, COST_start) 1.TRAIN_CD = prevalence(TRAIN) 2. C_i = build_clf(TRAIN, COST_start) 3. for (i=1; i<maxIterations; i++) 4. { 5. TEST_prob= classify(C_i, TEST) 6. TEST_CD = prevalence(TEST_prob) 7. COST = TEST_CD/TRAIN_CD 8.C_i = build_clf(TRAIN, COST) 9. } 10. return TEST_CD Figure 2 : Pseudo-code for the CDE-Iterate algorithm.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 297, |
| "end": 305, |
| "text": "Figure 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "To build a cost-sensitive classifier in the function build_clf, we tried a few ones and chose a fast logistic regression classifier.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "We did not find any open library where baseline quantification methods were implemented. We, therefore, shared all the algorithms, which we had programmed using the Python language, on the Github repository 1 . We believe that this library can help pool information on quantification.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Iterative Class Distribution Estimation", |
| "sec_num": "2.6" |
| }, |
| { |
| "text": "This section describes our experimental setup. It describes the datasets we use, the specific experiments we run and the classifier induction algorithm we employ.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Experiment Methodology", |
| "sec_num": "3" |
| }, |
| { |
| "text": "We present a simple experiment that illustrates the efficiency of iterative adjustment of the a priori probabilities.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "We use random sample generators from SkiKit Library to build artificial datasets of controlled size and complexity 2 . For each dataset we generate 10 ords with 10 features. Figure 3 exemplifies a dataset with two classes.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 174, |
| "end": 182, |
| "text": "Figure 3", |
| "ref_id": "FIGREF0" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "The initial prevalence for classes (p train (c 1 ) = p train (c 2 ) = 0.5). The total set randomly splits into two subsets: 25% training set, 75% test set. training set, the class distribution remains unchanged. For the test set, we vary prevalence value 0.05 to 0.95. For each prevalence value we generate a ferent test sets. Therefore, nineteen hundred of the following experimental design are applied.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "We used a Kullback-Leibler Divergence (KLD) tween the true class prevalence and the 2 http://scikitlearn.org/stable/modules/generated/sklearn.da sification.html erators from SkiKit-Learn Library to build artificial datasets of controlled size and dataset we generate 10,000 recexemplifies 2 features of es c 1 and c 2 was equal otal set randomly splits into two subsets: 25% training set, 75% test set. For the training set, the class distribution remains unchanged.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "prevalence value (c 1 ) from and TEST dataset items {0.5, 0.5} and TEST_CD = {0.1, 0.9} respectively (generated with 2 features).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "For each prevalence value we generate a hundred difnineteen hundred replications of the following experimental design are applied.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "Leibler Divergence (KLD) between the true class prevalence and the predicted class learn.org/stable/modules/generated/sklearn.datasets.make_clas prevalence as a quality evaluation metrics for quantif ers.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Simulations on Artificial Data", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "To evaluate the algorithms on the real data pated in the SemEval-2016 Task 4 called \"Sentiment Analysis in Twitter\". Its dataset consists of sages (aka observations) divided Task 4 consists of five subtasks, but w ed in subtasks D and E: tweet quantification according to a two-point scale and five These subtasks are evaluated topics, and the final result is counted as an average of evaluation measure out of all the topics 2016).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "The organizers provide a default split of the data into training, development and development tasets. The algorithms evaluation is performed these subsets. The training subset is used as a TRAIN set, development and development are used as a TEST set.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "Since observation x in this dataset is a message wri ten in a natural language, we first need to transform it to the vector representation X. Based on a study by and Sebastiani, 2015), we choose the following comp nents of the feature vector:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "\uf0b7 TFIDF for word n-grams with n 4", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "\uf0b7 TFIDF character n-grams where n 5.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "Feature vector is extracted with a We also perform data preprocessing terns (e.g. links, emoticons, numbers) w with their substitutes. For word n matization using WordNetLemmatizer.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "It is interesting to characterize messages using SentiWordNet library. For each token we obtain its polarity value from the SentiWordNet. First, we recognize the part of speech using tagger from the NLTK library cond, we get the SentiWordNet first polarity value for this token using the part of speech information.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "We used polarity values to extend vector represent tion of documents in two ways the polarity score as a sum of positive minus negative polarity values and add this feature to tor representation of a document. Second the sum of positive polarities and 3 http://scikitlearn.org/stable/modules/generated/sklearn.feature_extraction. text.TfidfVectorizer.html a quality evaluation metrics for quantifi-To evaluate the algorithms on the real data, we partici-2016 Task 4 called \"Sentiment Its dataset consists of Twitter mesdivided into several topics. Task 4 consists of five subtasks, but we only participat-D and E: tweet quantification according point scale and five-point scale, respectively. independently for different final result is counted as an average of evaluation measure out of all the topics (Nakov et al., default split of the data into training, development and development-time testing datasets. The algorithms evaluation is performed using raining subset is used as a TRAIN set, development and development-time testing subsets in this dataset is a message written in a natural language, we first need to transform it to . Based on a study by (Gao , we choose the following compograms with n varying from 1 to grams where n varies from 3 to extracted with a Scikit_Learn tool 3 . We also perform data preprocessing .Several text patlinks, emoticons, numbers) were replaced For word n-grams we apply lemmatization using WordNetLemmatizer.", |
| "cite_spans": [ |
| { |
| "start": 803, |
| "end": 817, |
| "text": "(Nakov et al.,", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "It is interesting to characterize messages using the SentiWordNet library. For each token x i in document X obtain its polarity value from the SentiWordNet. part of speech using a speech NLTK library (Bird et al., 2009) . Seget the SentiWordNet first polarity value for part of speech information. We used polarity values to extend vector representation of documents in two ways: first we simply calculate sum of positive minus a sum of negative polarity values and add this feature to the vecpresentation of a document. Second, we calculate sum of positive polarities and the sum of negative learn.org/stable/modules/generated/sklearn.feature_extraction. polarities and add these two features to the vector representation of a document.", |
| "cite_spans": [ |
| { |
| "start": 200, |
| "end": 219, |
| "text": "(Bird et al., 2009)", |
| "ref_id": "BIBREF1" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "The metrics that we use to evaluate the classifier performance are described in (Nakov et al., 2016) and are not described here.", |
| "cite_spans": [ |
| { |
| "start": 80, |
| "end": 100, |
| "text": "(Nakov et al., 2016)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Dataset", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "We apply six quantification methods mentioned above in Section 2: CC, PCC, ACC, PACC, EM, CDEIterate and compare them.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Experiment Results", |
| "sec_num": "4" |
| }, |
| { |
| "text": "First, we applied CC, PCC, ACC, PACC, EM and CDEIterate algorithms to generated data described in Section 3.1. Synthesized data allows us to perform a comparative analysis of these quantification methods with different amount of distribution drift.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Synthesized Data", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "In Figure3, which demonstrates the means and standard deviation values of the evaluation measure -Kullback-Leibler Divergence (KLD), each point is obtained by averaging over one hundred generated datasets with different prevalence. It is obvious from Figure 4 that the CDEIterate approach shows the lowest KLD mean values when a distribution drift is relatively large. A standard deviation value for the CDEIterate method remains the smallest one among all possible distribution drifts.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 251, |
| "end": 259, |
| "text": "Figure 4", |
| "ref_id": "FIGREF1" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Synthesized Data", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "On the contrary, the EM approach shows very unstable results. Sometimes the EM algorithm converges far from the real value. Its standard deviation displays the same unstable behavior.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Synthesized Data", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "For more careful consideration, let us show its functions in the logarithmic scale in Figure 5 . When distribution changes from the starting value p train (c) = 0.5 by less than 0.1, the simple methods like CC and PCC show better performance (lower KLD).", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 86, |
| "end": 94, |
| "text": "Figure 5", |
| "ref_id": "FIGREF2" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Synthesized Data", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "We noticed that CDEIterate methods sometimes converge to different values, if an algorithm starts iteration from a different starting point. To support this, we add the COST_start variable to the algorithm shown in Figure 2 . The first starting point is a priori probability distribution of a training set. Therefore, for the starting iteration we assume TEST_CD to equal TRAIN_CD. The second starting point is when TEST_CD is uniformly distributed. This case is labeled as CDEIterate_U. In the previous Section 4.1, these two starting points were actually the same. CDEIterate_U approach showed the best accuracy on the testing set among others with both five-point and two-point scales.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 215, |
| "end": 223, |
| "text": "Figure 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Test Data", |
| "sec_num": "4.2" |
| }, |
| { |
| "text": "SentiWordNet is usually regarded as an important source of information about word sentiment (Baccianella et al., 2010; Esuli and Sebastiani, 2006) . In our comparison, we add the sum of positive scores and the sum of negative scores of each word as two additional features to the feature vector. Only the first meaning, according to the recognized part of speech, was used. The quantification methods remain the same. The results provided in We explain this behavior as follows: simple algorithms cannot adjust to the whole singularity and such additional features increase dimension and, thereby, accuracy. In a more complex case, the classifier extracts information from features more efficiently. Additional information about polarity scores leads to algorithm overtraining. We can guess that, as tweets contain creative spelling and abbreviation common in Twitter (like \"lol\", not presented in SentiWordNet), the existence of character n-grams contains more specific information than polarity scores of selected, properly written words. Therefore, we exclude SentiWordNet features from the final feature vector.", |
| "cite_spans": [ |
| { |
| "start": 92, |
| "end": 118, |
| "text": "(Baccianella et al., 2010;", |
| "ref_id": "BIBREF0" |
| }, |
| { |
| "start": 119, |
| "end": 146, |
| "text": "Esuli and Sebastiani, 2006)", |
| "ref_id": "BIBREF3" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Test Data", |
| "sec_num": "4.2" |
| }, |
| { |
| "text": "The aim of this research was to perform comparative analysis of different approaches of state-of-the-art quantification techniques.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion and future work", |
| "sec_num": "5" |
| }, |
| { |
| "text": "For tweet quantification on a five-point scale (Subtask E) and a two-point scale (Subtask D), the best performance was demonstrated by the adopted iterative method proposed by (Xue and Weiss, 2009) , based on the iterative procedure with the cost-sensitive supervise learner. All the algorithms mentioned in the article, are available on the Github repository 4 .", |
| "cite_spans": [ |
| { |
| "start": 176, |
| "end": 197, |
| "text": "(Xue and Weiss, 2009)", |
| "ref_id": "BIBREF16" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion and future work", |
| "sec_num": "5" |
| }, |
| { |
| "text": "In our future work, we are planning to move in two directions. First, we plan to extend the vector of features used for representation of documents. Second, we want to add more quantification methods to our open source library.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion and future work", |
| "sec_num": "5" |
| }, |
| { |
| "text": "https://github.com/Arctickirillas/Rubrication", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "The reported study was funded by RFBR under research Project No. 16-06-00184 A.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Acknowledgments", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "SentiWordNet 3.0: An Enhanced Lexical Resource for Sentiment Analysis and Opinion Mining", |
| "authors": [ |
| { |
| "first": "Stefano", |
| "middle": [], |
| "last": "Baccianella", |
| "suffix": "" |
| }, |
| { |
| "first": "Andrea", |
| "middle": [], |
| "last": "Esuli", |
| "suffix": "" |
| }, |
| { |
| "first": "Fabrizio", |
| "middle": [], |
| "last": "Sebastiani", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "LREC", |
| "volume": "10", |
| "issue": "", |
| "pages": "2200--2204", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Stefano Baccianella, Andrea Esuli, and Fabrizio Sebastiani. 2010. SentiWordNet 3.0: An Enhanced Lex- ical Resource for Sentiment Analysis and Opinion Min- ing. In LREC, volume 10, pages 2200-2204.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Natural language processing with Python", |
| "authors": [ |
| { |
| "first": "Steven", |
| "middle": [], |
| "last": "Bird", |
| "suffix": "" |
| }, |
| { |
| "first": "Ewan", |
| "middle": [], |
| "last": "Klein", |
| "suffix": "" |
| }, |
| { |
| "first": "Edward", |
| "middle": [], |
| "last": "Loper", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Steven Bird, Ewan Klein, and Edward Loper. 2009. Natural language processing with Python. O'Reilly Media, Inc.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Estimating class priors in domain adaptation for word sense disambiguation", |
| "authors": [ |
| { |
| "first": "Yee", |
| "middle": [], |
| "last": "Seng Chan", |
| "suffix": "" |
| }, |
| { |
| "first": "Hwee Tou", |
| "middle": [], |
| "last": "Ng", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of the 21st International Conference on Computational Linguistics and the 44th", |
| "volume": "", |
| "issue": "", |
| "pages": "89--96", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Yee Seng Chan and Hwee Tou Ng. 2006. Estimating class priors in domain adaptation for word sense disam- biguation. In Proceedings of the 21st International Con- ference on Computational Linguistics and the 44th an- 4 https://github.com/Arctickirillas/Rubrication nual meeting of the Association for Computational Lin- guistics, pages 89-96. Association for Computational Linguistics.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Sentiwordnet: A publicly available lexical resource for opinion mining", |
| "authors": [ |
| { |
| "first": "Andrea", |
| "middle": [], |
| "last": "Esuli", |
| "suffix": "" |
| }, |
| { |
| "first": "Fabrizio", |
| "middle": [], |
| "last": "Sebastiani", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of LREC", |
| "volume": "6", |
| "issue": "", |
| "pages": "417--422", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Andrea Esuli and Fabrizio Sebastiani. 2006. Sentiwordnet: A publicly available lexical resource for opinion mining. In Proceedings of LREC, volume 6, pages 417-422. Citeseer.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Sentiment quantification. IEEE intelligent systems", |
| "authors": [ |
| { |
| "first": "Andrea", |
| "middle": [], |
| "last": "Esuli", |
| "suffix": "" |
| }, |
| { |
| "first": "Fabrizio", |
| "middle": [], |
| "last": "Sebastiani", |
| "suffix": "" |
| }, |
| { |
| "first": "Ahmed", |
| "middle": [ |
| "Abbasi" |
| ], |
| "last": "", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "", |
| "volume": "25", |
| "issue": "", |
| "pages": "72--79", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Andrea Esuli, Fabrizio Sebastiani, and Ahmed ABBASI. 2010. Sentiment quantification. IEEE intelli- gent systems, 25(4):72-79.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Counting positives accurately despite inaccurate classification", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Forman", |
| "suffix": "" |
| } |
| ], |
| "year": 2005, |
| "venue": "Machine Learning: ECML 2005", |
| "volume": "", |
| "issue": "", |
| "pages": "564--575", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George Forman. 2005. Counting positives accurately despite inaccurate classification. In Machine Learning: ECML 2005, pages 564-575. Springer. bibtex: for- man2005counting.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Quantifying counts and costs via classification", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Forman", |
| "suffix": "" |
| } |
| ], |
| "year": 2008, |
| "venue": "Data Mining and Knowledge Discovery", |
| "volume": "17", |
| "issue": "2", |
| "pages": "164--206", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George Forman. 2008. Quantifying counts and costs via classification. Data Mining and Knowledge Discovery, 17(2):164-206, June.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "Pragmatic text mining: minimizing human effort to quantify many issues in call logs", |
| "authors": [ |
| { |
| "first": "George", |
| "middle": [], |
| "last": "Forman", |
| "suffix": "" |
| }, |
| { |
| "first": "Evan", |
| "middle": [], |
| "last": "Kirshenbaum", |
| "suffix": "" |
| }, |
| { |
| "first": "Jaap", |
| "middle": [], |
| "last": "Suermondt", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining", |
| "volume": "", |
| "issue": "", |
| "pages": "852--861", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "George Forman, Evan Kirshenbaum, and Jaap Suermondt. 2006. Pragmatic text mining: minimizing human effort to quantify many issues in call logs. In Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 852-861. ACM.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Tweet Sentiment: From Classification to Quantification", |
| "authors": [ |
| { |
| "first": "Wei", |
| "middle": [], |
| "last": "Gao", |
| "suffix": "" |
| }, |
| { |
| "first": "Fabrizio", |
| "middle": [], |
| "last": "Sebastiani", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "Proceedings of the 2015 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2015", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Wei Gao and Fabrizio Sebastiani. 2015. Tweet Senti- ment: From Classification to Quantification. In Pro- ceedings of the 2015 IEEE/ACM International Confer- ence on Advances in Social Networks Analysis and Min- ing 2015, pages 97-104. ACM. bibtex: gao2015tweet.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "Cost-sensitive and modular land-cover classification based on posterior probability estimates", |
| "authors": [ |
| { |
| "first": "A", |
| "middle": [], |
| "last": "Guerrero-Curieses", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Alaiz-Rodriguez", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Cid-Sueiro", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "International Journal of Remote Sensing", |
| "volume": "30", |
| "issue": "22", |
| "pages": "5877--5899", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "A. Guerrero-Curieses, R. Alaiz-Rodriguez, and J. Cid- Sueiro. 2009. Cost-sensitive and modular land-cover classification based on posterior probability estimates. International Journal of Remote Sensing, 30(22):5877- 5899.", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "Classifier technology and the illusion of progress", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "David", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "Statistical science", |
| "volume": "21", |
| "issue": "1", |
| "pages": "1--14", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "David J. Hand and others. 2006. Classifier technology and the illusion of progress. Statistical science, 21(1):1- 14.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "Sentiment Analysis in Twitter", |
| "authors": [], |
| "year": 2016, |
| "venue": "Proceedings of the 10th International Workshop on Semantic Evaluation", |
| "volume": "4", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Task 4: Sentiment Analysis in Twitter. In Proceedings of the 10th International Workshop on Semantic Eval- uation (SemEval 2016), San Diego, California, June. Association for Computational Linguistics. bibtex: SemEval:2016:task4.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "Adjusting the outputs of a classifier to new a priori probabilities: a simple procedure", |
| "authors": [ |
| { |
| "first": "Marco", |
| "middle": [], |
| "last": "Saerens", |
| "suffix": "" |
| }, |
| { |
| "first": "Patrice", |
| "middle": [], |
| "last": "Latinne", |
| "suffix": "" |
| }, |
| { |
| "first": "Christine", |
| "middle": [], |
| "last": "Decaestecker", |
| "suffix": "" |
| } |
| ], |
| "year": 2002, |
| "venue": "Neural computation", |
| "volume": "14", |
| "issue": "1", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marco Saerens, Patrice Latinne, and Christine Decaestecker. 2002. Adjusting the outputs of a classifier to new a priori probabilities: a simple procedure. Neural computation, 14(1):21-41. bibtex: saerens2002adjusting.", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "Classification and quantification based on image analysis for sperm samples with uncertain damaged/intact cell proportions", |
| "authors": [ |
| { |
| "first": "Lidia", |
| "middle": [], |
| "last": "S\u00e1nchez", |
| "suffix": "" |
| }, |
| { |
| "first": "V\u00edctor", |
| "middle": [], |
| "last": "Gonz\u00e1lez", |
| "suffix": "" |
| }, |
| { |
| "first": "Enrique", |
| "middle": [], |
| "last": "Alegre", |
| "suffix": "" |
| }, |
| { |
| "first": "Roc\u00edo", |
| "middle": [], |
| "last": "Alaiz", |
| "suffix": "" |
| } |
| ], |
| "year": 2008, |
| "venue": "Image Analysis and Recognition", |
| "volume": "", |
| "issue": "", |
| "pages": "827--836", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lidia S\u00e1nchez, V\u00edctor Gonz\u00e1lez, Enrique Alegre, and Roc\u00edo Alaiz. 2008. Classification and quantification based on image analysis for sperm samples with uncer- tain damaged/intact cell proportions. In Image Analysis and Recognition, pages 827-836. Springer.", |
| "links": null |
| }, |
| "BIBREF15": { |
| "ref_id": "b15", |
| "title": "Network quantification despite biased labels", |
| "authors": [ |
| { |
| "first": "Lei", |
| "middle": [], |
| "last": "Tang", |
| "suffix": "" |
| }, |
| { |
| "first": "Huiji", |
| "middle": [], |
| "last": "Gao", |
| "suffix": "" |
| }, |
| { |
| "first": "Huan", |
| "middle": [], |
| "last": "Liu", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Proceedings of the Eighth Workshop on Mining and Learning with Graphs", |
| "volume": "", |
| "issue": "", |
| "pages": "147--154", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lei Tang, Huiji Gao, and Huan Liu. 2010. Network quantification despite biased labels. In Proceedings of the Eighth Workshop on Mining and Learning with Graphs, pages 147-154. ACM.", |
| "links": null |
| }, |
| "BIBREF16": { |
| "ref_id": "b16", |
| "title": "Quantification and semi-supervised classification methods for handling changes in class distribution", |
| "authors": [ |
| { |
| "first": "", |
| "middle": [], |
| "last": "Jack Chongjie Xue", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Gary", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Weiss", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Jack Chongjie Xue and Gary M Weiss. 2009. Quantifi- cation and semi-supervised classification methods for handling changes in class distribution. In Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 897-906. ACM. bibtex: xue2009quantification.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "FIGREF0": { |
| "text": "An example of TRAIN and TEST with TRAIN_CD = {0.5, 0.5} and TEST_CD spectively (generated with 2 fea", |
| "type_str": "figure", |
| "num": null, |
| "uris": null |
| }, |
| "FIGREF1": { |
| "text": "Mean and standard deviation values of Kullback-Leibler Divergence for different distribution drifts in the TEST set on the linear scale.", |
| "type_str": "figure", |
| "num": null, |
| "uris": null |
| }, |
| "FIGREF2": { |
| "text": "Mean and standard deviation values of Kullback-Leibler Divergence for different distribution drifts in the TEST set on the logarithmic scale.", |
| "type_str": "figure", |
| "num": null, |
| "uris": null |
| }, |
| "TABREF1": { |
| "content": "<table><tr><td>Method</td><td>Quantification accuracy measure</td></tr><tr><td>CC</td><td>0.868282929268</td></tr><tr><td>ACC</td><td>0.861784553862</td></tr><tr><td>PCC</td><td>1.05532269963</td></tr><tr><td>PACC</td><td>1.0731851762</td></tr><tr><td>EM</td><td>1.11319538187</td></tr><tr><td>CDEIterate</td><td>0.58872710467</td></tr><tr><td colspan=\"2\">CDEIterate_U 0.587811269105</td></tr></table>", |
| "type_str": "table", |
| "html": null, |
| "text": ", show that the new features increase quantification accuracy for CC, ACC, but surprisingly decrease it for PCC, PACC, EM, CDEIterate and CDEIterate-U. Comparison of methods on test sample with a fivepoint scale with additional SentiWordNet features (SemEval-2016 Task4 Subtask E).", |
| "num": null |
| } |
| } |
| } |
| } |
