Early search engines were Boolean in nature. A user gives a list of query terms, and the documents are marked as "relevant" or "non relevant" depending on whether or not they contain the terms in question. But for large document collections, just dividing the documents into two groups isn't much help. What if there are millions of documents in the "relevant" group? How is the user to decide which one to read first? The vector space model was introduced into information retrieval as early as the 1970s to address this problem. In the vector space model, documents are represented as points whose coordinates are determined by the terms contained in the document. Documents are queries are compared by taking the Euclidean distance between them, or more often the cosine of the angle in between them. But the mid-2000s, researchers had realized that there is more than similarity available in such search engines: one can use vector logic to model negation and disjunction as well. The logic in question was recognized as the same as the quantum logic used by Birkhoff and Von Neumann to describe quantum systems in the 1930s. Related Research Groups, Resources and Papers
|
Applications >