Analysis of network data to find related search queries and topics users are most interested in
Highlights
Instructor: Lada Adamic
Course: SI508 – Networks: Theory and Application (Fall’2008)
Teammates: Rohit L., & Jeff F.
What we did
Search queries entered by users are a valuable information source. They can be used as a form of indirect feedback from users about a service. We constructed a network model of a search log of a reference database publisher. In the network, nodes represent search term and edges represent relationships between queries. Search terms entered in one session were considered as related.
Our team developed a network model of the query log that can be used to provide relevant query suggestions to the user, using the data from the query. We wrote scripts that extracted the query, session id, and location from which the query was performed. This data was imported into Pajek (a network modeling tool), and we generated visualizations that can used by the reference database companies to understand user needs.