ACE: A Concept Extraction Approach using Linked Open Data

Given the increase in popularity of several social networks, numerous users tend to express themselves or reach out to their followers via online posts, normally in the form of microposts. Dealing with such data of short textual content can be quite intricate due to several factors such as misspellings, slang, emoticons, etc. In this paper we present an approach towards extracting several concepts from microposts, where the main challenge is to classify them into specific entity types. This will help in discovering knowledge from possible semistructured/unstructured data after taking into account several factors. In our approach we extend a state-of-the-art information extraction system which we call ACE, and make use of a dataset that is part of the Linked Open Data cloud, in order to improve the named entity extraction process.