Title | ||
---|---|---|
Social-Correlation Based Mutual Reinforcement for Short Text Classification and User Interest Tagging. |
Abstract | ||
---|---|---|
Short text such as micro-blog messages is becoming increasingly prevalent in China. Due to the sparseness of the features associated with short text, accurately classifying short text and tagging user interest have become important and challenging tasks. Many recent studies have focused on utilizing external data to address the data sparsity issue but fail to leverage the social-correlation which is expected to help improve the accuracy of short text classification. In this paper, we present a new method using a semi-supervised coupled mutual reinforcement framework based on social-correlation to simultaneously classify short text and tag user interest. Specifically, our method requires relatively few labeled examples to initialize the training process. More importantly, experimental results have demonstrated that our method can achieve 100% accuracy in classifying certain categories and significantly improve the accuracy of classifying the other categories. Meanwhile, the experiments show that our model is effective in user interest tagging. © Springer-Verlag 2013. |
Year | DOI | Venue |
---|---|---|
2013 | 10.1007/978-3-642-53914-5_38 | ADMA (1) |
Keywords | Field | DocType |
mutual reinforcement,short text classification,user interest tagging | Information retrieval,Computer science,Correlation,Artificial intelligence,Reinforcement,Machine learning | Conference |
Volume | Issue | ISSN |
8346 LNAI | PART 1 | 16113349 |
Citations | PageRank | References |
1 | 0.35 | 12 |
Authors | ||
2 |