ICTNET at Session Track TREC2014

Abstract

In this paper, we describe our solutions of the Session Track at TREC 2014. Our main idea is to re-rank the documents the official supplies as RL1. In order to get good results of the re-ranked documents, we implement the learning to rank model which needs to extract some features. We use the relevance judgments of Session Track TREC 2013 as training set this year and also we use it as testing set by 5- fold cross-validation. The rest of this paper is organized as follows. We detail our models in section 2. Section 3 describes our experiments, including our evaluation results. Conclusions are made in the last session.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 01, 2014
Accession Number
ADA618635

Entities

People

  • Guoxin Cui
  • Xiaoming Yu
  • Xueqi Cheng
  • Yuanhai Xue
  • Yue Liu

Organizations

  • Chinese Academy of Sciences

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Data Mining
  • Data Science
  • Feature Extraction
  • Feature Selection
  • Information Operations
  • Instructions
  • Judgment
  • Learning
  • Machine Learning
  • Maryland
  • Standards
  • Test And Evaluation
  • Training
  • Validation

Fields of Study

  • Computer science

Readers

  • Information Retrieval
  • Neural Network Machine Learning.