Can we use DBpedia for identifying Polysemy of words?

Question

I want to detect words with Polysemy in a given text in descending order. That the word that has multiple meanings should be in the top of the list, and words with least Polysemy should be in bottom.

In other words, I want a list of words with the number of "meanings" for each word (i.e. with highest Polysemy).

Is there a way to do it using DBpedia or any other resource in open data cloud?

Answer 1

First, I assume all NLP-related aspects are subtracted from your question.

DBpedia is considered as belonging to the Linguistic Linked Open Data Cloud (interactive SVG), but I'd suggest to try more specialized resources, e. g. RDF version of BabelNet.

One can't retrieve the whole list of English words with the number of meanings on the public endpoint.
Probably one could download a dump for local querying.

For relatively small sequences of words you could write something like this:

SELECT ?label (COUNT(?sense) AS ?number) WHERE {
     VALUES (?label) {("every"@en) ("drop"@en) ("of"@en) ("rain"@en) ("that"@en) ("fall"@en)}
     ?entries a lemon:LexicalEntry ;
              lemon:language "EN" ;
              lemon:sense ?sense ;
              rdfs:label ?label .
  #  ?sense lemon:reference/bn-lemon:synsetType "concept" .
} GROUP BY ?label ORDER BY DESC(?number)

Try it!

Results of the above query:

   label     number  
 ---------- -------- 
  "fall"         48  
  "drop"         33  
  "rain"          9  
  "that"          4  
  "every"         3  
  "of"            1  

"Senses" of "drop" in BabelNet 4.0.