Comparison of Learning Performance and Retrieval Performance for Support Vector Machines Based Relevance Feedback Document Retrieval

This paper presents a learning performance and a retrieval performance of an interactive document retrieval method, which is based on support vector machine(SVM). Some works have been done to apply classification learning like SVM to relevance feedback and obtained successful results. However they did not fully utilize characteristic of example distribution in document retrieval. We propose heuristics to bias document showing in order to take good learning performance and good retrieval performance of relevance feedback. This paper introduces two evaluation crietria. One criterion measures the learning performance and the other measures the retrieval performance. We compared a SVM-based system with our heuristic with conventional systems like Rocchio-based system and a SVM-based system without our heuristic by the introduced crietria. We could confirm the learning performance of our system outperformed other ones.