Course talk:CPSC522/Future Directions for Semantic Systems

Hi Samprity,

Great work with the page! I really like the selection of your two papers. I think you have also done a great job at explaining and illustrating the concepts (the fruit example). I guess a little more background on ontology search would have benefitted me more because this is an entirely new concept for me. Other than that, I think you have covered the topic in depth and structured it well too. I guess some more kind of general layman view either in abstract or introduction if you could add, the naive reader would be able to grasp the importance of the papers better. Hope this helps! :)

Warm regards,

Ritika

Thanks for the inputs! I will try to implement them.

Nice page. I wonder if you can have a section that compares OntoSearch and WI OntoSearch. There are some comparisons in the last paragraph but I think a clean and concise summary of difference would be helpful.

There was nothing explicitly mentioned in the papers. I can add some from my understanding.

Hi Samprity,

Very nice job and interesting topic. After reading, I have several questions.

How do they get the relationship with the apple, fruit, food thing. Is it pre-setted relationship? or this is systematic obtained knowledge? According to the content of the files over the internet, the system study and learn knowledge like data mining?

I am a little confused actually about the second paper. If a search engine considers more semantic, how can it filter the results instead of adding more related links? for example, when i search for fruit, it gets not only fruit, but also apples, grapes, is not it? I think what google is now doing not only just based on the keyword searched, but also based on related hot level or the users' interest. For example, if we search apple, then the first page is all about the company, but not the fruit. Or what I think is that actually this semantic system, would try to provide user a more efficient search mode, for example, they might be able to rank different results referencing the probability.

Hi Dandan
The relationship with the apple, fruit, food is just an example for explaining how the algorithm works. The second paper considers the structure of the ontology while searching. So it ranks and gives results that are similar to the input query and filters out ontologies which have lower resultant vector value. A 3 words input would give you a better precision in the results than a 1 word input. Also these papers were written in 2004 and 2005. So google's api for ontology search must have changed from back then.

Hi Samprity,

Nice to see your page regarding the knowledge about ontology search engine and its afterwards page of onto engine and it is well explained and helped me a lot in understanding the concepts of semantic webs. I am wondering whether there is any performance comparison or experiment regarding the two pages to show the incremental performance improvement for the two methods.

Regards Arthur Sun

Hi Arthur There is nothing explicit that I could find on their performance comparison. Current OntoSearch UI is completely different than the one mentioned in this paper. You can check it at http://www.ontosearch.com/. And WI OntoSearch isn't accessible. In the evaluation section- in my opinion the number of results for before filtering kind of deals with OntoSearch as it was based on simple keywords just like a normal google search engine. And after applying the algorithm ie in WI OntoSearch the number of results reduced as it filtered out irrelevant ontologies.

Hi Samprity,

Great page! I have a couple of confusions about your page. I did not get where do RDP and OWL come from. I thought that RDP might be a typo and RDF is the correct one. (According to the acronyms of its definition) I did not understand OWL. I think it needs to be explained more. I think that the background is not enough. I feel like the ontology search has not been explained and so I cannot get the motivation of the explained papers.

Let me know if you add more background. Then I can read again and understand better.

Thanks,

Bahare

Thanks for the feedback. Yes RDF was a typo. I have changed it. I will add more background for RDF and OWL. I had provided the motivation for Ontology Search Engine in the problem definition ie how keyword based search engines fall short for ontology search. I will still add more stuff there. Thanks for the inputs!