Unit 7 Reading Notes

IIR Chapter 9

·          The Relevance Feedback is the idea that the system involves the user’s feedback to refine the searching results.

·          Algorithms for implementing relevance feedback.

n   Rocchio Algorithm: incorporating relevance feedback information into the vector space model.

n   Naive Bayes probabilistic model

·          Relevance feedback can improve both recall (more effective) and precision.

·          Requirements for effective relevance feedback.

n   The user has to have sufficient knowledge to be able to make an initial query.

n   Relevant documents to be similar to each other.

·          Evaluating the effectiveness of relevance feedback

n   Start with an initial query q0 and to compute a precision-recall graph.

n   Use documents in the residual collection for the second round of evaluation.

·          Pseudo relevance feedback automates the manual part of relevance feedback, so that the user gets improved retrieval performance with- out an extended interaction.

·          Indirect relevance feed back uses indirect sources of evidence.

·          Implicit feedback is less reliable than explicit feedback, but is more useful than pseudo relevance feedback.

·          Three global methods for expanding a query: by simply aiding the user in doing so, by using a manual thesaurus, and through building a thesaurus automatically.

Reading: Improving the Effectiveness of Information Retrieval with Local Context Analysis

·          This paper proposes a new technique for automatic query expansion, called local context analysis, which selects expansion terms based on co-occurrence with the query terms within the top-ranked documents.

·          Existing techniques for automatic query expansion can be categorized as either global or local.

·          Local context analysis is a local technique, but it employs co-occurrence analysis, a primary tool for global techniques, for query expansion.

·          The metrics used by local context analysis for concept selection: co-occurrence metric, combining the degrees of co-occurrence with all query terms, differentiating rare and common query terms

·          Experimental results on a number of collections show that local context analysis is more effective than existing techniques.

Reading: A Study of Methods for Negative Relevance Feedback

·          This paper focuses on the analysis of negative relevance feedback. The Experiment results on several TREC collections show that language model based negative feedback methods are generally more effective than those based on vector-space models, and using multiple negative models is an effective heuristic for negative feedback.

·          General strategies with some variations for negative feedback: (1) SingleQuery: query modification strategy; (2) SingleNeg: score combination with a single negative query model; (3) MultiNeg: score combination with multiple negative query models.

·          Two heuristics to increase the robustness of using negative feedback information: Local Neighborhood and Global Neighborhood.