The following experiments shows online natural language processing experiments and applications, specifically related to semantic analysis.
The main experiments take advantage of a semantic network, examples of the semantic network and the path between two words in that network is shown in the figures to the right.
Experiments include: Words Semantic Relatedness and Similarity, Word-Phrase Semantic Similarity, and Semantic Compositionality in Context.
Experiments
Words semantic relatedness is a measurement of the degree to which two words are related.
The following online demo calculates two words semantic relatedness using the shorted path in a weighted semantic network. The weights are based on human annotators ratings of the relatedness of a specific path type.
For example, if you compute the semantic relatedness of: "Car" and "Vehicle" based on this demo, the result is that they are 91.5% semantically related.
This online demo evaluate semantic similarity of words and compositional phrases. The demo evaluates the semantic similarity of a word and a short sequence of two words. For example, in the word "contact" and the phrase "close interaction" are semantically similar.
This online demo returns the relations between words described by a set of prepositions and coordinations. For example, "Car" is related with the following set: (with, on, and, without, behind, at, of) to the word "Wheel"
This online demo returns the relations between words described by a set of predicates. For example, "Car" is related with the following set: (have, equip with, run on, fit with, be on, mount on, be with, have on...) to the word "Wheel"
This online demo evaluate the compositionality of phrases in context. the demo decides whether a phrase is used in its literal or figurative meaning in a given context. For example, the phrase "big picture" as in the context "Click here for a bigger picture" is used literally, whereas in the context "To solve this problem, you have to look at the big picture." it is used figuratively.