Identifying Authorities in Online Communities

Abstract

Several approaches have been proposed for the problem of identifying authoritative actors in online communities. However, the majority of existing methods suffer from one or more of the following limitations: (1) there is a lack of an automatic mechanism to formally discriminate between authoritative and nonauthoritative users. In fact, a common approach to authoritative user identification is to provide a ranked list of users expecting authorities to come first. A major problem of such an approach is where to stop reading the ranked list of users? How many users should be chosen as authoritative? (2) supervised learning approaches for authoritative user identification suffer from their dependency on the training data. The problem here is that labeled samples are more difficult, expensive, and time consuming to obtain than unlabeled ones. (3) several approaches rely on some user parameters to estimate an authority score. Detection accuracy of authoritative users can be seriously affected if incorrect values are used. In this paper, we propose a parameterless mixture model-based approach which is capable of addressing the three aforementioned issues in a single framework. In our approach, we first represent each user with a feature vector composed of information related to its social behaviour and activity in an online community. Next, we propose a statistical framework, based on the multivariate beta mixtures, in order to model the estimated set of feature vectors. The probability density function is therefore estimated and the beta component that corresponds to the most authoritative users is identified. The suitability of the proposed approach is illustrated on real data extracted from Stack Exchange question-answering network and Twitter.

Measuring user influence on Twitter: A survey

“…IDM-CTMP [78] ? Parameterless mixture [14] ? Table 1: Classification of measures in order of appearance in the article.…”

"Leadership in Action: How Top Hackers Behave" A Big-Data Approach with Text-Mining and Sentiment Analysis

“…If a community member has stayed for longer duration, the mutual trust demonstrated by fellow colleagues, assignment of duties, and their status improves [25]. Such behavior is also visible in community question-answering (CQA) forums [52] [53]. An earlier study defines the duration feature for forum analysis, merely with the help of the date of the first message posted [17].…”

“…Leaders are largely subject to moderate to high levels of credibility, because it signals highly dynamic community behavior [27] [51] [53]. A good amount of contribution indicates increased community activity and helps to build trust and reputation among other community members.…”

“…Often average message length for each message is used as a covariate [17] [38]. We find that relatively lengthy messages deliver more cognitive value and are far more important to the larger audience of the community -be it a hacker forum or an CQA such as StackOverflow [10] [52] [53]. Word counts and message lengths increase significantly across all types of information levels with rising depth of the messages in an online learning platform [39].…”

Security information sharing via Twitter: 'Heartbleed' as a case study

“…First, according to the data, there was no evidence that any particular group took on the role of authority. Secondly, the tendency of less knowledgeable groups to give advice on behaviour (promote password change when this was not the first step to a solution) suggests that establishing a voice of authority, representing the most knowledgeable experts, may be helpful to raise awareness and suggest action (see recent work on identifying authoritative actors by online communities by Bouguessa and Romdhane, 2015). And third, given the connections that marketing and entrepreneurs appear to foster (at least in this paper's dataset), these may also be suitable target groups to disseminate information, in addition to the press.…”