论文信息 - Using Cohesion-Model for Story Link Detection System

Using Cohesion-Model for Story Link Detection System

Summary Cohesion-Model is a new Story Link Detection (SLD) System that is inspired by the relevance model of TDT. Task in hand is to detect whether given two documents are linked. Each document is expanded using corresponding relevant documents. Each term in the relevant document is weighted according to the cohesion factor. The two models then are compared using the modified fractional similarity. Performance of this model shows distinct improvement when compared with most effective Link Detection System, which measures similarity between the given two documents using cosine method. The model is also compared with a system using modified fractional method without building Cohesion-Model. The experimental results show the performance of Cohesion-Model tested with TDT4 data and proves the effectiveness of the new model.

Saswati Mukherjee

[1] Yiming Yang,et al. Improving text categorization methods for event tracking , 2000, SIGIR '00.

[2] Victor Lavrenko,et al. A Generative Theory of Relevance , 2008, The Information Retrieval Series.

[3] Saswati Mukherjee,et al. An Improved Feature Selection using Maximized Signal to Noise Ratio Technique for TC , 2006, Third International Conference on Information Technology: New Generations (ITNG'06).

[4] Thorsten Brants,et al. Optimizing Story Link Detection is not Equivalent to Optimizing New Event Detection , 2003, ACL.

[5] James Allan,et al. Introduction to topic detection and tracking , 2002 .

[6] James Allan,et al. Using Soundex Codes for Indexing Names in ASR Documents , 2004, HLT-NAACL 2004.

[7] James Allan,et al. Relevance models for topic detection and tracking , 2002 .

[8] James Allan,et al. Capturing term dependencies using a language model based on sentence trees , 2002, CIKM '02.

[9] Maria Elena Figueroa,et al. Communication for social change: an integrated model for measuring the process and its outcomes. , 2002 .

[10] James Allan,et al. UMass at TDT 2000 , 2000 .

[11] Yiming Yang,et al. Topic Detection and Tracking Pilot Study Final Report , 1998 .

[12] Thorsten Brants,et al. Multiple Similarity Measures and Source-Pair Information in Story Link Detection , 2004, HLT-NAACL.