Collaborative personalised dynamic faceted search

Siripinyo Chantamunee

Information retrieval systems are facing challenges due to the overwhelming volume of available information online. It leads to the need of search features that have the capability to provide relevant information for searchers. Dynamic faceted search has been one of the potential tools to provide a list of multiple facets for searchers to filter their contents. However, being a dynamic system, some irrelevant or unimportant facets could be produced. To develop an effective dynamic faceted search, personalised facet selection is an important mechanism to create an appropriate personalised facet list. Most current systems have derived the searchers' interests from their own profiles. However, interests from the past may not be adequate to predict current interest due to human information-seeking behaviour. Incorporating current interests from other people's opinions to predict the interests of individual person is an alternative way to develop personalisation which is called Collaborative approach. This research aims to investigate the incorporation of a Collaborative approach to personalise facet selection. This study introduces the Artificial Neural Network (ANN)-based collaborative personalisation architecture framework and Relation-aware Collaborative AutoEncoder model (RCAE) with embedding methodology for modelling and predicting the interests in multiple facets. The study showed that incorporating collaborative approach into the proposed framework for facet selection is capable to enhance the performance of personalisation model in facet selection in comparison to the state-of-the-art techniques.

Collaborative personalised dynamic faceted search

Files and links (1)

Abstract

Details

Metrics