Dialogue – Driven Intranet Search Suma Adindla School of Computer Science & Electronic Engineering 8th LANGUAGE & COMPUTATION DAY May LAC 09

Motivating Example Imagine you could interact with a university intranet search engine as follows: User : Head of department System: Which department are you looking for? User: Computer science System: The head of the computer science department is Dr. Maria Fasli. Her contact details are as follows... Do you want any further information? User: How do I get to her office? System: The quickest route from the University information point is as follows May LAC 09

Short queries in most cases (often just keywords rather than questions) Unstructured data to start with Explicit domain knowledge Wide range of possible user queries 23 May LAC 09 Problem Overview

30 October LAC 09 Table 1. Research Context

Possible Solution 23 May LAC 09 Impose a dialogue manager to assist a user in the navigation process Turn document collection into usable knowledge source Employs automatically extracted domain knowledge

23 May LAC 09 Research Questions to Address 1. How can a dialogue system be incorporated into the search process to provide the user with a more natural language interface? 2. Can such a dialogue system on an intranet provide the user with more relevant information and offer a better user experience than a standard search engine?

System Architecture 23 May LAC 09 User request clarifications suggestions Dialogue Manager Search Engine Domain Model response Query Analysis

Online query processing Match a user query against the knowledge base Imposed dialogue system helps a user in the navigation of results 30 October LAC 09 Document Processing Identifying Triplets Extracting Entities and Facts Query Analysis (Question type) Pre- processing Query mapping Process Dialogue Manager Domain Knowledge User

Extraction process 23 May LAC 09 Sentence Detection Parts of Speech Tagging Parsing Named Entity Recognition Reference Resolution Documents

Evaluation Parameters Number of interaction steps (dialogue length) Time taken to process a user query Precision and recall (success rate of retrieved results) User satisfaction 23 May LAC 09

Future challenges Extracting useful knowledge from the document collection Knowledge representation Interactive dialogue manager with some domain knowledge Natural language interface design 23 May LAC 09

Thank you 30 October LAC 09