We address the problem of the categorization of short texts, like those posted by users on social networks and microblogging platforms. We specifically focus on Twitter. Since short texts do not provide sufficient word occurrences, and they often contain abbreviations and acronyms, traditional classification methods such as "Bag-of-Words" have limitations. Our proposed method enriches the original text with a new set of words, to add more semantic value by using information extracted from webpages of the same temporal context. Then we use those words to query Wikipedia, as an external knowledge base, with the final goal to categorize the original text using a predefined set of Wikipedia categories. We also present a first experimental evaluation that confirms the effectiveness of the algorithm design and implementation choices, highlighting some critical issues with short texts.
