Stack Overflow系统的特征融合答案推荐策略

FEATURE INTEGRATION ANSWER RECOMMENDATION STRATEGY FOR STACK OVERFLOW SYSTEM

针对Stack Overflow系统中用户寻找问题答案效率低的问题,提出一种基于标题相似度、描述相似度、标签相似度、语义相似度的特征融合答案推荐策略(FIARS)。从Stack Overflow网站中抽取“问题与答案”语料集,对答案进行去重处理,建立问题索引和问题对应的答案集索引;采用余弦相似度计算新问题与语料库中问题在标题、标签、问题描述等维度上的相似度,并构建语义模型计算语义相似度;基于这些相似度筛选出最佳的“问题与答案”候选集并把答案推荐给用户。为了验证策略的可行性和有效性,使用Stack Overflow真实数据集进行分析实验,实验结果表明该策略能够较大地提高答案推荐的准确率。

Aiming at the problem of low efficiency of searching for answers in Stack Overflow system,this paper proposed a feature integrated answer recommendation strategy(FIARS) based on title similarity,description similarity,tag similarity and semantic similarity.We extracted the "question and answer" corpus from the Stack Overflow website,eliminated duplication and built the question and the answer set index.The cosine similarity was used to calculate the similarity between the new problem and the corpus in terms of title,tag and problem description,and the semantic model was constructed to calculate the semantic similarity.Based on these similarities,the best candidate sets of "questions and answers" were selected,and the answers were recommended to users.In order to verify the feasibility and effectiveness of the strategy,we used real data set from Stack Overflow to analyze the experiment.The experimental results show that the strategy can greatly improve the accuracy of answer recommendation.