Bulletin boards are, for many years, a well-known and established way of providing information to different groups of people. We tried to present an improved internet-based form of a bulletin board where we had to first inform ourselves about the basic natural language processing tasks. Among other things, we performed tokenization of the published content, lemmatization of the obtained tokens and also built a structure of semantically similar words in a non-relational database. We also classified the texts using a naive Bayesian classifier, thus allowing the contextual matching of the posts. We successfully tested the implemented search and match systems on an internship problem domain in the shape of a web service based on the supplied content from educational institutions as well as companies and organizations.
展开▼
