{"title":"Graph-Based Methods for Language Processing and Information Retrieval","authors":"Dragomir R. Radev","doi":"10.1109/SLT.2006.326781","DOIUrl":"https://doi.org/10.1109/SLT.2006.326781","url":null,"abstract":"Summary form only given. A number of problems in information retrieval and natural language processing can be approached using graph theory. Some representative examples in IR include Brin and Page's Pagerank and Kleinberg's HITS for document ranking using graph-based random walk models. In NLP, one could mention Pang and Lee's work on sentiment analysis using graph min- cuts, Mihalcea's work on word sense disambiguation, Zhu et al.'s label propagation algorithms, Toutanova et al.'s prepositional attachment algorithm, and McDonald et al.'s dependency parsing algorithm using minimum spanning trees. In this talk I will quickly summarize three graph-based algorithms developed recently at the University of Michigan: (a) lexrank, a method for multidocument summarization based on random walks on lexical centrality graphs, (b) TUMBL, a generic method using bipartite graphs for semi-supervised learning, and (c) biased lexrank, a semi-supervised technique for passage ranking for information retrieval and discuss the applicability of such techniques to other problems in Natural Language Processing and Information Retrieval.","PeriodicalId":74811,"journal":{"name":"SLT ... : ... IEEE Workshop on Spoken Language Technology : proceedings. IEEE Workshop on Spoken Language Technology","volume":"6 1","pages":"4"},"PeriodicalIF":0.0,"publicationDate":"2006-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89351600","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Model Adaptation for Dialog Act Tagging","authors":"Gökhan Tür, Ümit Güz, Dilek Z. Hakkani-Tür","doi":"10.1109/SLT.2006.326825","DOIUrl":"https://doi.org/10.1109/SLT.2006.326825","url":null,"abstract":"In this paper, we analyze the effect of model adaptation for dialog act tagging. The goal of adaptation is to improve the performance of the tagger using out-of-domain data or models. Dialog act tagging aims to provide a basis for further discourse analysis and understanding in conversational speech. In this study we used the ICSI meeting corpus with high-level meeting recognition dialog act (MRDA) tags, that is, question, statement, backchannel, disruptions, and floor grabbers/holders. We performed controlled adaptation experiments using the Switchboard (SWBD) corpus with SWBD-DAMSL tags as the out-of-domain corpus. Our results indicate that we can achieve significantly better dialog act tagging by automatically selecting a subset of the Switchboard corpus and combining the confidences obtained by both in-domain and out-of-domain models via logistic regression, especially when the in-domain data is limited.","PeriodicalId":74811,"journal":{"name":"SLT ... : ... IEEE Workshop on Spoken Language Technology : proceedings. IEEE Workshop on Spoken Language Technology","volume":"204 1","pages":"94-97"},"PeriodicalIF":0.0,"publicationDate":"2006-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77023227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}