On the difficulty of clustering company tweets

Twitter is a new successful technology of the Web 2.0 genre which is used by millions of people and companies to publish brief messages ("tweets") with the purpose of sharing experiences and/or opinions about a product or service. Due to the huge amount of information available in this type of technology, there is a clear need for new systems that can mine these messages in order to derive information about the collective thinking of twitterers (e.g. for opinion or sentiment analysis). Tweet analysis is a very important task because comments, opinions, suggestions, complaints can be used as marketing strategies or for determining information on a company's reputation. For this purpose, it is necessary to establish whether a tweet refers to a company or not, which is not a straightforward keyword search process as there may be multiple contexts in which a name can be used. The aim of this work is to present and compare a number of different approaches based on clustering that determine whether a given tweet refers to a particular company or not. For this purpose, we have used an enriching methodology in order to improve the representation of tweets and as a consequence the performance of the clustering company tweets task. The obtained results are promising and highlight the difficulty of this task.