Sentiment Analysis and Opinion Mining Opinion Mining

The invention relates to a comprehensive evaluation trust-based social network recommendation model construction method The method comprises the following steps of: preliminarily confirming a trust relationship between a recommended user and a neighbor user; establishing evaluation similarity trust between the recommended user and the neighbor user; establishing social relationship similarity trust between the recommended user and the neighbor user; establishing PageRank trust between the recommended user and the neighbor user; and synthesizing the results of above steps to obtain a comprehensive evaluation trust value between the users According to the method provided by the invention, the aim of correctly quantizing the trust relationship between the recommended user and the neighbor user in a social network is realized Through the quantized trust relationship, the disadvantage that the traditional recommendation models which provide recommendation services by utilizing user evaluation similarities are low in recommendation correctness is improved, and the requirements for personalized recommendation of recommended users in the social network are satisfied to the greatest extent

Comprehensive evaluation trust-based social network recommendation model construction method

The invention provides a user association method across website based on username similarity. The steps comprise: 1) filtering characters in a plurality of usernames, just reserving English letters and numbers; 2) finding out characteristics of the processed usernames, and obtaining self-information values of the characteristics, according to the self-information values, obtaining self-information vectors; 3) according to the self-information vectors, obtaining similarity among the plurality of usernames, if the similarity is larger than a given threshold value [tau], determining that the plurality of usernames belong to a same user. Through the similarity among the plurality of usernames, the method determines whether the usernames belong to the same user, and accounts belonging to the same user on different websites can be associated.

User association method across website based on username similarity

The invention relates to a cross-social network user identity recognition method and system based on machine learning. The method comprises the following steps of: acquiring cross-network user information; the text analysis precision user identity alignment annotation and user information feature extraction module is used; and based on construction of an alignment user candidate set of user attribute information and a user social relationship, based on construction and parameter learning of an accurate user identity alignment model spliced by the user attribute information, user release content and user social representation characteristics, providing a query service of cross-network user identity alignment and constructing a cross-network integrated user portrait. The cross-network user identity alignment is realized through the accurate identity alignment model formed by splicing the characteristics of the user attention relationships of the attribute sets of different network users,and a more detailed user portrait is constructed.

Cross-social network user identity recognition method and system based on machine learning

The invention provides a recognition method and system of accounts in batch registration. The method comprises the following steps: extracting the first user registration information from all registered users in the preset time quantum, wherein the user registration information comprises the unique identification information of a first user and nickname field information of the first user; computing the editing distance between the nickname field information of the first user and the nickname field information of all other registered users in the preset time quantum one by one; computing the editing similarity between the nickname field information of the first user and the nickname field information of all other registered users one by one according to the editing distance; computing theaverage similarity of the nickname field information of the first user and at least one nickname field information with the highest similarity based on the editing similarity between the nickname field information of the first user and the nickname field information of all other registered users; and judging that the first user is in the batch registration risk when the average similarity is greater than the preset threshold.

Recognition method and system of accounts in batch registration

The present invention discloses an integrated data mining method for similarity of accounts on social network sites. The method can be used for monitoring online public opinions and solves the identification problem that the same user has a plurality of accounts on the social network sites. The method disclosed by the invention comprehensively considers three factors which affect the integrated similarity of the accounts of the social network sites: personal attribute, inter-behavior and content, and determines the weight of the similarity of each factor in calculation of the integrated similarity by using a training sample. Compared to the prior art, the method disclosed by the invention has the technical advantages that (1) the method provides quantitative, reliable and comprehensive reference for identifying the plurality of the accounts on the social network sites of the same user and is suitable for automatic processing by a computer under a big data environment; and (2) the weight of the similarity of the each factor in the calculation of the integrated similarity is determined by adopting the training sample, so that the consistence of the result and a manual processing result can be maintained.

Integrated data mining method for similarity of accounts on social network sites

In the big data era, massive news reports about the latest events are being published on the Web. To thoroughly understand an event, we have to read massive reports and keep clues in mind, which is very difficult and usually results in a one-sided interpretation. In this paper, we propose a multi-document summarization approach which summarizes reports of a particular social or political event automatically and comprehensively. To speed up summarization, a pre-summarization approach is introduced to condense each report to a sub-summary, which can reduce the scale of subsequent processing. As an event should be told in chronological order, a timeline is introduced to organize and aggregate event-relevant sub-summaries. With each day's sub-summaries, a key phrase extraction algorithm is used to cluster them into topics and generate a meaningful label for each topic. Finally, a selection criterion is introduced to select relevant and novel sentences for each topic. We perform experiments on a large-scale news dataset, with about 10 million reports collected from news sites. An empirical study shows that our system is feasible under large scale environment. An evaluation on effectiveness shows that it is favoured by users.

Summarizing Events from Massive News Reports on the Web

Opinions on the news reports are very useful for people to understand an event, while the vast availability of such opinions becomes overwhelming to people when there is too much to digest. In this paper, we present an approach on summarizing event opinions for news reports. Firstly, we search relevant reports by submitting event-describing keywords, and for each report, we extract main textual content and separate subjective opinions from objective facts. Secondly, we propose a subjectivity detection approach to identify the opinion sentences. Thirdly, in order to summarize opinions, we propose a digest approach to list and compare standpoints by classifying opinions into positive/negative camps according to their sentiment polarities. In each camp, besides salient opinions, we also identify and present opinion holders (e.g. People, organizations, and countries), which contribute to a better understanding of their stances and relations. Lastly, we develop a system to display the opinion summarization of the event that is searched. An evaluation on effectiveness shows that it is favored by users.

http://ieeexplore.ieee.org/iel7/7311006/7311794/07311967.pdf

Chang Liu

Papers

Integrated data mining method for similarity of accounts on social network sites

Summarizing Events from Massive News Reports on the Web

Summarizing Event Opinions for News Reports