9Months Ago .. i was still remembering all the talks about Semantic web and Web 3.0 .. all those definitions , some were true and some were false.

it was the time i was searching for a Graduation project lots of ideas were popping out and lots of fears of spending a year of work in a traditional thing.

i was proceeding my internship @ CMIC lab and in fact i felt my self annoying keeping asking every one about his opinion in some ideas that popped out in my head . hoping that in a way or anohter we could  get into an idea that seemed to be very cool.

after less than a month and more than a dozen of chats & meetings and more than hundreds of articles read , the idea was clear . it was a  Semantic search engine that is capable of answering Questions in a semantic way not by Text matching answers like most of traditional Search engines do. something the user could search in with terms like ” give me all android phones with capacitive touch ” or “what is the salary of cristiano ronaldo ”  or “presidential candidates in US presidential elections 2010 ”

the idea was a little vague and it’s domains of search were unclear and unknown ,  this  one day i was asking a Developer about some  keywords i could search about in the domain . then he told me an answer that seemed to be off the topic at the first time , however it was very helpful afterwards . ” why not searching in the domain of semantic web , it’s a fresh research domain that alot of people are searching in”  he said , i did a little research in semantic web at that time  , but that answer made me rethink of semantic web. a little time until i found that it’s a very promising domain with a millions of dumps of semantically annotated data , disambiguation handled  that makes  every project  tasks  easy for the researchers in the domain of data mining and information retrieval . This makes me think now it would be very nice for them to shift Into  semantic web , that would end their suffer from finding corpus free from noise  in multi languages or preprocessing dumps and categorizing them .

The idea became clear and we were  3 in the team and now came the time for finding a professor to accept this project as our graduation project , it was a tiring phase  we contacted literally  11 professors  , some were uninterested and others liked the idea but didn’t want to involve in something they have no background in , until came the day before the deadline of proposal submission that time , when a 12th professor of our contact list accepted the proposal and willingly wanted to mentor our idea .

After loads and loads of exploration,  we found some promising papers  done by a  professor named kaufmann and others related to a workshop for semantic web “QALD-1”  , We had an apartment and began development  , we made a process , we had our source controller bug tracker and a little server for hosting the database server .

I don’t remember the number of meetings we had  they were nearly every day for 10 hours  less or more a day , but I’m sure we exceeded  20 meetings a month .

Long story short , yesterday was the project defense , we were presenting  ” WEET-IT ”  (know – it )   our semantic knowledge engine , weet-it  handles  natural language queries in lots of domains  , letting the using ask  natural language free questions  for example like :

  • What is the birth place of barack obama  ?
  • What is the area of Egypt  ?
  • What are the countries that the river nile flows through ?
  • What is the profession of the wife of barack obama ?
  • What is the birth place of the children of hosni mubarak ?
  • What is the position of Messi ?

Previewing answers in form of profiles for the entities returned

As well as  questions that needs special visualization other than direct answers  like

  • Compare between Egypt and United states ?
  • What is the relation between Egypt and aboutreka ?

Our bench marks were very promising relatively to the evaluation of  “aqua log ” and ” FREYA ” the top question answering systems  in the QALD  workshops for semantic web

We believe that our results are  competitive to  EVI  the famous  Question answering application .

In the next days , we will be searching for fund for hosting on servers  as well as  some expansion and refinement

Our next steps would be adding more interesting statistics and datasets to our data base as well as exploration of new algorithms to enhance some search queries , aiming to handle subjective queries like   ” what is the best android phone with capacitive touch ”   or  ” tell me what people think of saddam hussien ? ”

” presidential candidates of Egypt ? ”  , previewing answers of charts and statistics elaborating people sentiment and reviews about those products and entities

ps : we are searching for funding for hosting on larger servers and expansions issues , for interesting organizations you can contact me   or any of the team members .

Advertisements