论文部分内容阅读
[目的/意义]提出一种以情感加权算法和朴素贝叶斯算法为基础的组合分类模型(SWNB模型),旨在对中文微博话题的立场进行判别。[方法/过程]该模型首先通过给定的复杂句模型对微博进行简化,然后依据情感规则得到情感权值,提取微博中与话题相关的实体并进行优化,进而将微博分为包含立场和未表明立场(NONE)两类;再对包含立场的微博提取特征词,利用朴素贝叶斯算法将其立场判别为支持(FAVOR)或反对(AGAINST)。[结果/结论 ]实验结果表明,本模型有较好的立场判别精度,并能同时有效地处理中文复杂句式、话题相关评价对象以及上下文语境等复杂情形。
[Purpose / Significance] A combinatorial classification model (SWNB model) based on emotion-weighted algorithm and Naive Bayes algorithm is proposed to discriminate the position of Chinese Weibo topic. [Method / Process] This model first simplifies the Weibo by a given complex sentence model, then obtains the emotional weight according to the emotion rule, extracts and optimizes the topic-related entities in the Weibo, Position and unspecified position (NONE). Secondly, feature words are extracted from micro-blogs that contain positions and their positions are judged as FAVOR or AGAINST using Naive Bayes algorithm. [Results / Conclusions] The experimental results show that this model has better discrimination accuracy of position and can simultaneously handle complex situations such as Chinese complex sentences, topic related evaluation objects and contexts.