Abstract | ||
---|---|---|
Questions in Community question answering (CQA) consisting of some labeled questions and numerous unlabeled questions are so complex and irregular. Therefore, question classification in CQA has become the research hotspot in recent years. In this paper, we propose to classify the questions in CQA through the label propagation algorithm (LPA) based on the concept of graph, where nodes represent the labeled and unlabeled sample questions and edges represent the distance between the sample questions, through the node label propagation to realize question classification. Experiments on corpuses from “Baidu Knows”, the accuracy in question classification through the LPA is not only higher than that through the KNN algorithm and SVM algorithm that have applied the labeled samples, but also higher than that through the SVM-based Bootstrapping algorithm that has utilized the labeled and unlabeled samples. |
Year | Venue | Field |
---|---|---|
2015 | ICSI | k-nearest neighbors algorithm,Graph,Question answering,Pattern recognition,Label propagation,Bootstrapping,Computer science,Support vector machine,Artificial intelligence,Machine learning |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
3 | 4 |