Towards a characteristics-aware document search engine

The increasing volume, heterogeneity, and redundancy of the Web create a novel challenge for search engines in which, target documents must satisfy some characteristics. It is increasingly important because there are more and more types of web pages on the Internet nowadays. Current web search engines are fundamentally incapable of addressing the user need because keywords can not express characteristics of target pages. Another alternative is to use vertical search engines, but they can only cover a few popular niches. Thus, we propose Forward Search to empower users with an engine to express not only topics but also characteristics of pages in their queries. Expected results are documents ranked by both topical and characteristic relevance. Creating Forward Search to have the focus of a vertical engine and the flexibility of web search presents many novel challenges. First, we must represent a document with novel information to support querying for characteristics. Second, we must index both keywords and named entities to quickly locate relevant pages of the target characteristics during query time. Third, we have to design a realistic method to acquire user input about the characteristics of target pages and retrieve documents ranked by both topical and characteristic relevance. Finally, since our system redefines relevance in traditional web search, we must rethink the user interface. This thesis comes with a web-based demonstration of our Forward Search at http://crow.cs.illinois.edu:8080 and five open-source code repositories at https://github.com/forward-uiuc.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2018

Author dc:creator

• Pham, Long T

Contributors dc:contributor

• Chang, Kevin C.C.

• full text search, entity search, document retrieval, characteristics aware

Statement dc:rights

• Copyright 2018 Long Pham

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/101716

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F101716