1. A Generation Model to Unify Topic Relevance and Lexicon-based Sentiment for Opinion Retrieval Min Zhang, Xinyao Ye Tsinghua University SIGIR 2008 2009. 01. 14. Summarized & presented by Jung-Yeon Yang IDS Lab.

2. Copyright  2008 by CEBT Introduction  A growing interest in finding out people’s opinions from web data Product survey Advertisement analysis Political opinion polls TREC started a special track on blog data in 2006 – blog opinion retrieval – It has been the track that has the most participants in 2007  This paper is focused on the problem of searching opinions over general topics 2

3. Copyright  2008 by CEBT Related Work  The popular opinion identification approaches Text classification Lexicon-based sentiment analysis Opinion retrieval  Opinion retrieval To find the sentimental relevant documents according to a user’s query  One of the key problems How to combine opinion score with relevance score of each document for ranking 3

4. Copyright  2008 by CEBT Related Work (cont.)  Topic-relevance search is carried out by using relevance ranking(e.g. TF*IDF ranking)  Ad hoc solutions of combining relevance ranking and opinion detection result 2 steps: rank with relevance, then re-rank with sentiment score Most existing approaches use a linear combination α *Score rel + β *Score opn 4

5. Copyright  2008 by CEBT Backgrounds  Statistical Language Model (LM) A probability distribution over word sequences – p(“ Today is Wednesday ”)  0.001 – p(“ Today Wednesday is ”)  0.0000000000001 – Unigram : P(w1,w2,w3) = P(w1)*P(w2)*P(w3) – Bigram : P(w1,w2,w3) = P(w1)*P(w2|w1)*P(w3|w2) Can also be regarded as a probabilistic mechanism for “generating” text, thus also called a “generative” model  LM allows us to answer questions like: Given that we see “ John ” and “ feels ”, how likely will we see “ happy ” as opposed to “ habit ” as the next word? (speech recognition) Given that we observe “baseball” three times and “game” once in a news article, how likely is it about “sports”? (text categorization, information retrieval) 5

6. Copyright  2008 by CEBT Backgrounds (cont.)  The notion of relevance 6 Relevance  (Rep(q), Rep(d)) Similarity P(r=1|q,d) r  {0,1} Probability of Relevance P(d  q) or P(q  d) Probabilistic inference Different rep & similarity Vector space model Prob. distr. model … Generative Model Regression Model Classical prob. Model Doc generation Query generation LM approach Prob. concept space model Different inference system Inference network model

7. Copyright  2008 by CEBT Backgrounds (cont.)  Retrieval as Language Model Estimation Document ranking based on query likelihood Retrieval problem  Estimation of p(w i |d) Smoothing is an important issue – Problem : If tf = 0, then p(w i |d) = 0 – Smoothing methods try to Discount the probability of words seen in a document Re-allocate the extra probability so that unseen words will have a non-zero probability – Most use a reference model(collection language model) to discriminate unseen words 7 Document language model Discounted LM estimate Collection language model

8. Copyright  2008 by CEBT Generation Model for Opinion Retrieval  The Generation Model To find both sentimental and relevant documents with ranks  Topic Relevance Ranking  Opinion Generation Model and Ranking  Ranking function of generation model for opinion retrieval 8

9. Copyright  2008 by CEBT The Proposed Generation Model  Document generation model how well the document d “fits” the particular query q p(d |q) ∝ p(q |d)p(d)  In opinion retrieval, p(d |q, s ) Users’ information need is restricted to only an opinionate subset of the relevant documents This subset is characterized by sentiment expressions s towards topic q 9

10. Copyright  2008 by CEBT The Proposed Generation Model (cont.)  In this work, discuss lexicon-based sentiment analysis Assume – The latent variable s is estimated with a pre-constructed bag-of-word sentiment thesaurus – All sentiment words s i are uniformly distributed  The final generation model 10  quadratic relationship

11. Copyright  2008 by CEBT Topic Relevance Ranking I rel (d,q)  The Binary Independent Retrieval (BIR) model is one of the most famous ones in this branch Heuristic ranking function BM25 – TREC tests have shown this to be the best of the known probabilistic weighting schemes 11 IDF(w)

12. Copyright  2008 by CEBT Opinion Generation Model I op (d,q,s)  I op (d,q,s) focus on the problem that given query q, how probably a document d generates a sentiment expression s. Sparseness problem  smoothing Jelinek-mercer smoothing is applied. 12

13. Copyright  2008 by CEBT Opinion Generation Model I op (d,q,s)  Use the co-occurrence of s i and q inside d within a window W as the ranking measure of P ml (s i |d,q) 13

14. Copyright  2008 by CEBT Ranking function of generation model for opinion retrieval  The final ranking function  To reduce the impact of unbalance between #(sentiment words) and #(query terms)  logarithm normalization 14  Only use the topic relevance

15. Copyright  2008 by CEBT Experimental Setup  Data set TREC blog 06 & 07 100,649 blogs during 2.5 month  Strategy : find top 1000 relevant documents, then re-rank the list with proposed model  Models General linear combination Proposed generation model with smoothing Proposed generation model with smoothing and normalization 15

16. Copyright  2008 by CEBT Experimental Setup (cont.)  Sentimental Lexicons 16 Thesaurus nameSizeDesc. 1HowNet4621English translation of pos/neg Chinese words from HowNet 2WordNet7426Selected words from WordNet with seeds 3Intersection14131 ∩ 2 4Union10634 1 ∪ 2 5General Inquirer3642All words in the positive and negative categories 6SentiWordNet3133Words with a positive or negative score above 0.6

17. Copyright  2008 by CEBT Expetiment Results 17

18. Copyright  2008 by CEBT Conclusion  Proposed a formal generation opinion retrieval model Topic relevance & sentimental scores are integrated with quadratic comb.  Opinion generation ranking functions are derived  Discussed the roles of the sentiment lexicon and the matching window  It is a general model for opinion retrieval Use the domain-independent lexicons No assumption has been made on the nature of blog-structure text 18

19. Copyright  2008 by CEBT My opinions  Good points proposed the opinion retrieval model and ranking function Do various experiments  Lacking points Just find documents that has some opinions Don’t know what kinds of opinions in the documents Don’t use the sentimental polarity of words 19