Technical Questions Answering System

来自Data Management Lab
跳转至: 导航搜索

Overview

This is a system that answering technical questions. In this system, we use a two step pipeline:

At the first stage, we use some traditional information retrieval methods to select candidate documents;

At the second state, we use a technical-related knowledge graph to re-ranking the candidates;

Our two-step pipeline guarantees both precision and efficiency. The technical knowledge graph we construct covers full series products of Microsoft and is of high quality. Our re-ranking approach outperforms main-stream search engine and other state-of-art methods. And the index-based random walk approach speeds up the online search to a large extent. In the future, we are to further evaluate our method on other specific retrieval systems such as news, research papers recommendation.

All the technical documents are crawled from http://support.microsoft.com/en-us/kb.

Publication

Efficiently Answering Technical Questions --- A Knowledge Graph Approach

Shuo Yang, Lei Zou, Zhongyuan Wang, Jun Yan, Jirong Wen, AAAI 2017

People

  • Shuo Yang (Peking University)
  • Lei Zou (Peking University)
  • Zhongyuan Wang (Microsoft Asia)
  • Jun Yan (Microsoft Asia)
  • Jirong Wen (Renmin University & Beijing Key Laboratory of Big Data Management and Analysis Methods)