Finding Structured and Unstructured Features to Improve the Search Result of Complex Question

Wardani, Dewi Wisnu and Lu, Wen Hsiang (2009) Finding Structured and Unstructured Features to Improve the Search Result of Complex Question. Masters thesis, National Cheng Kung University.

[img] PDF - Published Version
Download (827Kb)

    Abstract

    -Recently, search engine got challenge deal with such a natural language questions. Sometimes, these questions are complex questions. A complex question is a question that consists several clauses, several intentions or need long answer. In this work we proposed that finding structured features and unstructured features of questions and using structured data and unstructured data could improve the search result of complex questions. According to those, we will use two approaches, IR approach and structured retrieval, QA template. Our framework consists of three parts. Question analysis, Resource Discovery and Analysis The Relevant Answer. In Question Analysis we used a few assumptions, and tried to find structured and unstructured features of the questions. Structured feature refers to Structured data and unstructured feature refers to unstructured data. In the resource discovery we integrated structured data (relational database) and unstructured data (webpage) to take the advantaged of two kinds of data to improve and reach the relevant answer. We will find the best top fragments from context of the webpage In the Relevant Answer part, we made a score matching between the result from structured data and unstructured data, then finally used QA template to reformulate the question. In the experiment result, it shows that using structured feature and unstructured feature and using both structured and unstructured data, using approach IR and QA template could improve the search result of complex questions.

    Item Type: Thesis (Masters)
    Subjects: T Technology > T Technology (General)
    Divisions: Fakultas Matematika dan Ilmu Pengetahuan Alam > Informatika
    Depositing User: mr admin 007
    Date Deposited: 21 May 2013 23:29
    Last Modified: 21 May 2013 23:29
    URI: https://eprints.uns.ac.id/id/eprint/823

    Actions (login required)

    View Item