面向查询意图歧义性的多样化检索模型研究  

Diversity Search on Query Intent of Ambiguity

作　　者：桂思思 张晓娟 GUI Si-si;ZHANG Xiao-juan(College of Information Management,Nanjing Agricultural University,Nanjing 210095,China;Faculty of Education,Southwest University,Chongqing 400115,China)

机构地区：[1]南京农业大学信息管理学院,江苏南京210095 [2]西南大学教育学部,重庆400715

出　　处：《情报科学》2021年第12期39-45,共7页Information Science

基　　金：国家社会科学基金青年项目“面向学术搜索的查询意图研究”(19CTQ023)。

摘　　要：【目的/意义】查询意图歧义性对检索模型提出了挑战。针对查询意图歧义性程度,探讨了基于歧义程度的多样化检索模型的检索效果。【方法/过程】将查询意图歧义性程度的表示方式分为序数变量或连续变量两种方式,在此基础上,提出了基于三种排序策略的面向序数变量查询意图歧义性的多样化检索模型、基于查询重构的面向连续变量查询意图歧义性的多样化检索模型,从而使得检索结果列表同时具有较高的覆盖率与新颖性。【结果/结论】在公开数据集上,四个检索效果测评指标α-nDCG@5、α-nDCG@10、α-nDCG@20及NRBP@20表明,本文提出的多样化检索模型优于基准实验,且获取准确的查询子主题能有效提升检索效果。【创新/局限】区分了查询意图歧义性程度的两种表示方式,据此提出并验证了面向查询意图歧义性程度的多样化检索模型;然而限于实验运行复杂程度,生成初始检索结果列表数据略少。【Purpose/significance】Query intent of ambiguity challenges retrieval models.This paper discusses the effectiveness of diversity search for dealing query intent with different ambiguities.【Method/process】This paper first argues that the query ambiguity can be measured by ordinal variables or continuous variables.Later,we propose a diversity search model to deal with ordinal query ambiguity through three rerank strategies and a diversity search model to deal with continuous query ambiguity through query reformulation.Finally,a diversified search result list is returned with both high coverage and novelty.【Result/conclusion】On a open access dataset,α-nDCG@5,α-nDCG@10,α-nDCG@20 and NRBP@20 show that,our proposed models outperform baseline and gathering accurate subtopics can effectively improve the effectiveness of the retrieval model.【Innovation/limitation】Our diversified retrieval models for dealing with ordinal or continuous query are proposed and validated.However,due to the experiment complexity,the initial retrieval result list is slightly small.

关 键 词：查询意图 歧义性 多样化检索 重排序 歧义性表示 

分 类 号：G254.9[文化科学—图书馆学]