Keyphrase Extraction Based on Multi-Feature

• Published in: 2019 International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI) pp. 208 - 213
• Main Authors: Dong, Hui, Wan, Jianyi, Wan, Zhongying
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2019
• Subjects: Data mining; Feature extraction; High frequency; Information filters; Keyphrase extraction; Multi-feature fusion; Probability; Thematic information; Web pages; Word position
• DOI: 10.1109/MLBDBI48998.2019.00047

The keyphrase extraction algorithm can be used in many fields such as information retrieval, recommendation system, text categorization, etc. The efficient keyphrase extraction method can quickly summarize the core content of the text. In order to improve the accuracy of extracting keyphrase, we propose a text keyphrase extraction method based on multi-feature. First, construct word graph by using the word co-occurrence relationship; Secondly, count the number of sentences in which the target word appears in the sentence, calculate the sum of the reciprocals of the positions of the words in the document, add the word span factor to filter the high frequency noise data, and extract the subject information by using the LDA model; Finally, integrate these characteristics and conduct a random walk on the word graph until the score becomes stable, rank the nodes according to the score and select appropriate words as keyphrase. We conducted experiments on three data sets SemEval2010, KDD and WWW. By studying the selection of various parameters of our algorithm model and comparing with various algorithms, the results show that the quality of keyphrase extraction in our model is improved.