Binyang Li

CLUSM: An Unsupervised Model for Microblog Sentiment Analysis Incorporating Link Information

Microblog has become a popular platform for people to share their ideas, information and opinions. In addition to textual content data, social relations and user behaviors in microblog provide us additional link information, which can be used to improve the performance of sentiment analysis. However, traditional sentiment analysis approaches either focus on the plain text, or make simple use of links without distinguishing different effects of different types of links. As a result, the performance of sentiment analysis on microblog can not achieve obvious improvement. In this paper, we are the first to divide the links between microblogs into three classes. We further propose an unsupervised model called Content and Link Unsupervised Sentiment Model (CLUSM). CLUSM focuses on microblog sentiment analysis by incorporating the above three types of links. Comprehensive experiments were conducted to investigate the performance of our method. Experimental results showed that our proposed model outperformed the state of the art.

Mainstream media behavior analysis on Twitter: a case study on UK general election

With the development of social media tools such as Facebook and Twitter, mainstream media organizations including newspapers and TV media have played an active role in engaging with their audience and strengthening their influence on the recently emerged platforms. In this paper, we analyze the behavior of mainstream media on Twitter and study how they exert their influence to shape public opinion during the UKs 2010 General Election. We first propose an empirical measure to quantify mainstream media bias based on sentiment analysis and show that it correlates better with the actual political bias in the UK media than the pure quantitative measures based on media coverage of various political parties. We then compare the information diffusion patterns from different categories of sources. We found that while mainstream media is good at seeding prominent information cascades, its role in shaping public opinion is being challenged by journalists since tweets from them are more likely to be retweeted and they spread faster and have longer lifespan compared to tweets from mainstream media. Moreover, the political bias of the journalists is a good indicator of the actual election results.

The Role of Physical Location in Our Online Social Networks

One of the most important properties of social networking sites is its reachability – no physical location constraint. In addition, all social networking sites allow us to search people with common interests, so we can find friends anywhere in the world easier than ever. With the help of social media, it seems that expaning our social networks is physical location independent. Motivated by the above observations, we study the role of physical location in social media. If physical location is no longer a barrier and physical interaction can be ignored, then our online social networks should have the following characteristics: (1) A number of our friends are from different places in the world other than the places that we have been; (2) A number of our friends are not from our physical social circles – they are not our colleagues, not our high school friends, etc.

A Chinese Sentence Compression Method for Opinion Mining

The Chinese sentences in news articles are usually very long, which set up obstacles for further opinion mining steps. Sentence compression is the task of producing a brief summary at the sentence level. Conventional compression methods do not distinguish the opinionated information from factual information in each sentence. In this paper, we propose a weakly supervised Chinese sentence compression method which aiming at eliminating the negligible factual parts and preserving the core opinionated parts of the sentence. No parallel corpus is needed during the compression. Experiments that involve both automatic evaluations and human subjective evaluations validate that the proposed method is effective in finding the desired parts from the long Chinese sentences.

Web Information Mining and Decision Support Platform for the Modern Service Industry

This demonstration presents an intelligent information platform MODEST. MODEST will provide enterprises with the services of retrieving news from websites, extracting commercial information, exploring customers’ opinions, and analyzing collaborative/competitive social networks. In this way, enterprises can improve the competitive abilities and facilitate potential collaboration activities. At the meanwhile, MODEST can also help governments to acquire information about one single company or the entire board timely, and make prompt strategies for better support. Currently, MODEST is applied to the pillar industries of Hong Kong, including innovative finance, modem logistics, information technology, etc.

An efficient approach for sentence-based opinion retrieval

Recently, there is a growing interest in sharing personal opinions on the Web, such as product reviews, economic analysis, political polls, etc. Therefore, opinion retrieval, which targets to retrieve documents expressing opinions or comments about the query, has become more and more popular. A typical method for opinion retrieval is document-based and each document is assigned a relevant score and an opinionated score, respectively. Then the documents are ranking based on a combination of the two scores. In this method, however, the document is split into bag-of-word, and the association between the opinion and its corresponding target is broken. In an extreme case, a relevant document full of irrelevant opinions will also be retrieved. In this paper, we propose a sentence-based approach since opinions are always expressed in one sentence where the association between an opinion and its corresponding target is maintained. We assign an individual score to each sentence rather than assign an overall score to the document directly. Moreover, we consider the effectiveness of different positions of sentences in documents to further capture the structural information. Compared with document-based approaches, experimental results on our own dataset show that our approach has achieved significant improvement.

Summarizing and extracting online public opinion from blog search results

As more and more people are willing to publish their attitudes and feelings in blogs, how to provide an efficient way to summarize and extract public opinion in blogosphere has become a major concern for both compute science researchers and sociologist. Different from existing literatures on opinion retrieval and summarization, the major issue of online public opinion monitoring is to find out people’s typical opinions and their corresponding distributions on the Web. We observe that blog search results could provide a very useful source for topic-coherent and authoritative opinions of the given query word. In this paper, a lexicon based method is proposed to enrich the representation of blog search results and a spectral clustering algorithm is introduced to partition blog search results into opinion groups, which help us to find out opinion distributions on the Web. A mutual reinforcement random walk model is proposed to rank result items and extract key sentiment words simultaneously, which facilitates user to quickly get the typical opinions of a given topic. Extensive experiments with different query words were conducted based on a real world blog search engine and the experiments results verify the efficiency and effectiveness of our proposed model and methods.

UIR-PKU: Twitter-OpinMiner System for Sentiment Analysis in Twitter at SemEval 2015

Microblogs are considered as We-Media information with many real-time opinions. This paper presents a Twitter-OpinMiner system for Twitter sentiment analysis evaluation at SemEval 2015. Our approach stems from two different angles: topic detection for discovering the sentiment distribution on different topics and sentiment analysis based on a variety of features. Moreover, we also implemented intra-sentence discourse relations for polarity identification. We divided the discourse relations into 4 predefined categories, including continuation, contrast, condition, and cause. These relations could facilitate us to eliminate polarity ambiguities in compound sentences where both positive and negative sentiments are appearing. Based on the SemEval 2014 and SemEval 2015 Twitter sentiment analysis task datasets, the experimental results show that the performance of Twitter-OpinMiner could effectively recognize opinionated messages and identify the polarities.

Pests Hidden in Your Fans: An Effective Approach for Opinion Leader Discovery

With the development of Web 2.0, people would like to share opinions on the Web, which are very helpful for other users to make decisions. Especially, some users have more powerful influence to other members of a community, group, or society, and their advice, opinions, and views are more valuable. We call these people opinion leaders. The study of opinion leader discovery from the social media is meaningful because it could help users to understand influential user behavior, and trace vital information diffusion of an e-society, even on-line ecology. However, existing approaches focus on linkage-based methods without considering the pests who have relationship with the potential opinion leader but carrying opposite opinions. In an extreme case, an opinion leader might be mistakenly identified according to his richer relationships with the pests. In this paper, we start from explaining the definition of opinion leader, and take into consideration of the user profile and posts’ opinions instead of using structural information (linkage) only. As such, those pests carrying opposite opinions could be gotten rid of from the social network, which could further improve the effectiveness of discovering opinion leaders. To evaluate the performance of our approach, we made experiments based on the Tweets data, and the results showed that our proposed approach could achieve 8% improvement compared with the linkage-based approach.

An effective approach for topic-specific opinion summarization

Topic-specific opinion summarization (TOS) plays an important role in helping users digest online opinions, which targets to extract a summary of opinion expressions specified by a query, i.e. topic-specific opinionated information (TOI). A fundamental problem in TOS is how to effectively represent the TOI of an opinion so that salient opinions can be summarized to meet users preference. Existing approaches for TOS are either limited by the mismatch between topic-specific information and its corresponding opinionated information or lack of ability to measure opinionated information associated with different topics, which in turn affect the performance seriously. In this paper, we represent TOI by word pair and propose a weighting scheme to measure word pair. Then, we integrate word pair into a random walk model for opinionated sentence ranking and adopt MMR method for summarization. Experimental results showed that salient opinion expressions were effectively weighted and significant improvement achieved for TOS.