Today, the Web is one of the world’s largest databases. Not surprisingly, much of the data is unstructured text (e.g., newsfeeds), the majority of which is user generated and difficult to manipulate or aggregate. Traditionally, search engines today rely on user-provided keywords. However, users must still comb the reported results to find the information they seek.
The primary objective in the Go-OLAP project is to recognize users’ search intent and provide information in real-time from web data sources. Initially, we will focus on simple examples, such as, “display everything about Barack Obama” or “list all presidents.”
In order to solve these types of queries, our Go-OLAP.info prototype will need to specifically address three critical problems. First, extract entities and relations from a large corpus of news data. Second, recognize users’ search intent. And lastly, display aggregate extracted data in a simple, fast and in an intuitive way.
The Go-OLAP project is lead by Alexander Löser.