Segmental analysis and evaluation of user focused search process

TY - CONF

T1 - Segmental analysis and evaluation of user focused search process

AU - Hendahewa, Chathra

AU - Shah, Chirag

PY - 2013

Y1 - 2013

N2 - In general, IR systems assist searchers by predicting or assuming what could be useful for their information needs by providing query suggestions or pseudo-relevance feedback. Most of these approaches are based on analyzing information objects (documents, queries) seen or used in the past and then proposing other related objects that may be relevant. Such approaches often ignore the underlying process of information seeking that guides how a searcher performs during information seeking episode, thus forgoing opportunities for making process-based recommendations. In order to address this, we propose a search process-based analysis of discovering different segments, which leads to analyzing different search action based features and evaluating the search performance for each stage. Further, we propose a query recommendation strategy to improve the search performance of each low performing user for each stage, which shows that the proposed overall model yields effective search performance improvements above 90% in most cases. This could lead to better recommendations and optimizations within each segment in order to enhance the overall search performance of a user.

AB - In general, IR systems assist searchers by predicting or assuming what could be useful for their information needs by providing query suggestions or pseudo-relevance feedback. Most of these approaches are based on analyzing information objects (documents, queries) seen or used in the past and then proposing other related objects that may be relevant. Such approaches often ignore the underlying process of information seeking that guides how a searcher performs during information seeking episode, thus forgoing opportunities for making process-based recommendations. In order to address this, we propose a search process-based analysis of discovering different segments, which leads to analyzing different search action based features and evaluating the search performance for each stage. Further, we propose a query recommendation strategy to improve the search performance of each low performing user for each stage, which shows that the proposed overall model yields effective search performance improvements above 90% in most cases. This could lead to better recommendations and optimizations within each segment in order to enhance the overall search performance of a user.

KW - Evaluation

KW - Exploratory search

KW - Sequence Analysis

KW - Time Series Analysis

UR - http://www.scopus.com/inward/record.url?scp=84899424956&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84899424956&partnerID=8YFLogxK

U2 - https://doi.org/10.1109/ICMLA.2013.59

DO - https://doi.org/10.1109/ICMLA.2013.59

M3 - Paper

SP - 291

EP - 294

T2 - 2013 12th International Conference on Machine Learning and Applications, ICMLA 2013

Y2 - 4 December 2013 through 7 December 2013

ER -