Bootstrapping Websites for Classification of Organization Names on Twitter

There has been a growing interest in monitoring the social media presence of companies for improved marketing. Many public APIs are available for tapping into the data, and there are companies that will collect all posts related to a given set of keywords. But with so much data, who is to say that all of the posts are relevant, especially when so many company and product names are highly ambiguous? In the context of the WePS Task 2, we aim to reduce noise by collecting only the relevant tweets about a company given the company's website and set of Twitter data. In a real world situation, any company who wanted to identify tweets about themselves could provide a short list of labeled tweets and use this as a base set for training data. Given that for this task we were given a large list of companies with no such training data, it would have been unrealistic to create such data for each company. We chose to use the company's website as surrogate training data. Because the websi...

• Type: Conference
• Year: 2010
• Where: CLEF
• Authors: Paul Kalmar
• Keywords: CLEF 2010, Company's Website, Information Technology, Many Public Apis, Training Data