Yes, you can use any well-defined ontology-based similarity measure and WordNet.

Precisely, I have submitted some papers during th last weeks with the novel state-of-the-art measures on WordNet, but I cannot provide them until be accepted.

  I suggest to use WordNet 3.1 and the Lin' measure with the intrinsic IC model of Seco et al. I provided you below the references of both, and a very good survey of Sebastien Harispe on the topic.

(1) Lin, D. (1998). An information-theoretic definition of similarity. In Proceedings of the 15th International Conference on Machine Learning, (Vol. 98, pp. 296–304). Madison, WI.

(2) Seco, N., Veale, T., & Hayes, J. (2004). An intrinsic information content metric for semantic similarity in WordNet. In R. López de Mántaras & L. Saitta (Eds.), Proceedings of the16th European Conference on Artificial Intelligence (ECAI) (Vol. 16, pp. 1089–1094). Valencia, Spain: IOS Press.

  You can also visit the Similarity portal of Ted Pedersen below. If you don´t want to code tehm, you could find some libraries in the Web.

http://wn-similarity.sourceforge.net/

Regards,

Juan Lastra-Díaz

 Here is the survey of Harispe on similarity and relatedness measures.

(1) Harispe, S., Ranwez, S., Janaqi, S., & Montmain, J. (2013). Semantic Measures for the Comparison of Units of Language, Concepts or Instances from Text and Knowledge Base Analysis. arXiv Preprint arXiv: Retrieved from http://arxiv.org/abs/1310.1285

I suggest to you follow 3 libraries

1-ws4j-1.0.1(َApi for Java)

2-NLTK (Python)

3- Wordnet:Similarity(in Perl language)

as relation ship is based on the is-a or  of-a type in nature, in that case best way to defined the relationship using ontology. as retrieval of result will be efficient and precise.  

Before a good answer can be given, try to decide your definition of similarity. How are two terms or words related? Exact similarity or synonymy is better defined and hence easier to computer. But what about similarity in one context and not in other? [Is DBPedia the right basis for context and hence disambiguation?] What about partial similarity or relatedness? What is the model in which you can consider matching and mapping?  To give you an analogy, see a relatively broader model of similarity called "semantic proximity" in http://dl.acm.org/citation.cfm?id=765855 . It requires you to consider the domain and context, the model of representation, whether you are limited to schema level information or are also looking at instances.

Now you may need to define the model for your need (which is not the database as in the reference above). There is a rich literature on semantic relationships...may want to check it.

On the other hand, current view on ontology alignment (and the benchmarks used for a popular evaluation referred to OAEI) offers a very narrow types of similarities/relationships, and hence has a limited value.

Maybe our paper on "Analysis of word co-occurrence in human literature for supporting semantic correspondence discovery" can help.

Regards, Mario

Conference Paper Analysis of word co-occurrence in human literature for suppo...

The IBM Watson "Concept Expansion Service API" might also be interesting for you ... you find it here: http://www.ibm.com/smarterplanet/us/en/ibmwatson/developercloud/concept-expansion.html

Hi there, in recent work of ours (see attachment) we have also used ws4j, suggested by Kazem above.

Conference Paper Monitoring Term Drift Based on Semantic Consistency in an Ev...

The first algorithm specifically designed for Short Text Semantic Similarity can be found here:

http://www.aaai.org/Papers/FLAIRS/2004/Flairs04-139.pdf

With a more extensive treatment here:

http://ants.iis.sinica.edu.tw/3BkMJ9lTeWXTSrrvNoKNFDxRm3zFwRR/55/Sentence%20Similarity%20Based%20on%20Semantic%20Nets%20and%20corpus%20statistics.pdf

This paper should give you some further insight on interpreting performance of Short Text Semantic Similarity measures:

https://www.researchgate.net/publication/262411640_A_new_benchmark_dataset_with_production_methodology_for_short_text_semantic_similarity_algorithms

"Short Texts" are of approximately sentence length. The *Sem group set a challenge on "Semantic Text Similarity" I analysed their datasets and found the text lengths to vary from 3 to over 60 words. You can find them at:

http://alt.qcri.org/semeval2015/task2/

Article A new benchmark dataset with production methodology for shor...

Dear Aniruddh,

I suggest to use SML toolkit.

Good luck.

http://www.semantic-measures-library.org/