Semantic Similarity and Relatedness

In order to provide different semantic relatedness measures to perform experiments, We implemented different types of structure-based methods by using WordNet and distributional semantic models by using Wikipedia.

Explicit Semantic Analysis

Explicit Semantic Model (ESA) was implemented to calculate the semantic relatedness scores between natural language texts. ESA is a distributional model that uses Wikipedia articles to build the distributional vector.

A RESTful service and a GUI are provided to use it.

Cross-Lingual Explicit Semantic Analysis

As Wikipedia contains articles in different languages, ESA model can provide the relatedness scores across different languages. Currently, our implementation supports English (en), German (de), Spanish (es), and Dutch (nl).

A RESTful service and a GUI are provided to use it.

WordNet-based Similarity

Since last two decades, several WordNet-based similarity methods have been proposed. We provided a RESTful service of eight different similarity measures by using WS4J (WordNet Similarity for Java) API.

A simple GUI is also provided to use it.