Enhancing Topic Detection in Twitter Using the Crowdsourcing Process

A decade ago, the crowdsourcing term was first coined and used to represent a method for expressing the wisdom of the crowd in accomplishing tasks that need human intelligence rather than machines and can be more efficiently accomplished time and financial wise using the crowd rather than indoor experts. This crowdsourcing process mainly contains four modules: designing incentives, then collecting, aggregating and verifying received information. The expert discovery module can be added to reduce the cost and enhance reliability and accuracy. The crowdsourcing process is used in this work to harness the mental ability of reliable Internet users around the globe and to improve the knowledge discovery techniques over social media; especially Twitter. The main objective is to improve the quality of the Twitter Exemplar-based topic detection system. The feedback from the crowd is utilized to adjust weights of the cosine similarity function deployed in the Exemplar-based topic detection algorithm. Testing the system using the Football Association Cup (FA Cup) dataset, it is found that the crowdsourcing has achieved a constant increase in the topic recall (by up to 15%), term precision (by up to 4%) and term recall (by up to 3%). Therefore, the new weights succeeded in increasing the three measures of topic quality significantly.