Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Copy of 1
Transcript of Copy of 1
The automatic crawler for Arabic content
Why using ontology based structure instead of a simple database querying schema?
Sparqueling the ontology to get answers
Creating the Qur'anic ontology
Here we had 2 approaches to go with
Relation detection crawler algorithm
Main relation norms
Found 2 main relation norms:
(n1, n2) have 0 words between
verb location is (n1-1) or (n2+1).
(n1, n2) have 1 word between
verb location is (n1+1).
Focus is on semantics, instances are optional.
Incremental, high re-usability.
Ontology axioms are used to specify meaning.
Reasoners are used to infer new information.
Comments are defined as a part of the ontology.
No query lock-in, queries have no restrictions.
Very easy to evolve and maintain, just start plugging in your concepts.
Focus is on data, instances are a must.
Often starts from scratch.
Database constraints are used to ensure integrity.
SQL engines are used for querying, views and data integrity.
Data dictionary is defined in a separate artifact.
Locked to specific queries per database, restricted to the schema.
Hard to maintain and expand, ETL tools to help.
Name of Prophet who was swallowed by the Fish
What is the punishment of the people of Noah?
The manual approach
Building the ontology by pure human effort, no algorithm or programing code used.
Less scientifically comprehensive but more accurate relations.
The automatic approach
Programming the process of eliciting triplet relations from the Qur'anic text.
Not very promising results but more scientifically comprehensive.
What is the problem?
Automatic Knowledge Extraction.
Semantic Search and Question Answering.
Users trying to find information in the Arabic texts have to enter exact words in search engines, which will only retrieve sentences that surround the words entered by the user.
Retrieved results are groups of texts that surround the questions words.
Not practical to users looking for answers to certain questions who don't know the exact words mentioned in the Qur'an or if they misspell their query.
Why the Qur'an?
Because it's considered the main information source for 1.5 billions Muslims.
Current approaches can't answer questions they only find exact word matches.
Proposed algorithm to extract relations from Arabic content
Nouns have relations between them.
Threshold of 0 or one word at most.
Relation is often a verb.
Example on case 1.1
Example on case 2
Example on case 1.2
A short demonstration...
A short demonstration...
What Animals are forbidden to eat?
M.Sc in CS thesis seminar (1)
Using Semantic Approaches to Answer Arabic Questions
from the Holy Qur'an.
Student: Hashim A. J. Shmaisani
Supervisor: Dr. Samir Al-Tartir
Faculty of Information Technology
Department of Computer Science
Current problem scenarios and approaches
Currently, users trying to find information in the Qur'an have to enter exact words in search engines, which will only retrieve sentences that surround the words entered by the user. This is not very practical to users looking for answers to certain questions who don't know the exact words mentioned in the Qur'an, misspell their query or even just looking for semantics around a certain concept in the Qur'an.
Current approaches to solve this problem
Current approaches reside on retrieving data from mass, unstructured and non-semantic content, using a an exact syntax match approach.
To enhance retrieval some researchers used datamining techniques, others used some word pre-processing techniques to enhance retrieval, but none reached the level of actually using semantic based structure to enhance retrieval or knowledge enferencing from Arabic content.
Developing an approach to semantically organize and structure Arabic content using semantic relation based structure in order to get better results for search, information retrieval and question answering proposes.
Conclusions and upcoming work
For future work we are aiming to create or modify a rule based stemmer for the Arabic language in order to enhance the time and space complexity of the crawler.
We are currently working on the semantic layer of the proposed model, which will provide synonyms to the user provided query in order to enhance the chances to retrieve an answer.
Question analysis layer
This will be the final layer of our proposed model, this level will infer the targeted domain in the ontology to be searched and the targeted classes and relations in order to retrieve the desired answer(ontology instance).
Lahsen Abouenour, on the improvement of passage retrieval in Arabic question/answering (Q/A) systems, proceedings of the 16th international conference on natural language processing and information systems, Springer-Verlag Berlin, Heidelberg, 2011.
Wafa N Bdour, Natheer K. Gharaibeh, Development of yes/no Arabic question answering system, international journal of artificial intelligence and applications Vol.4, January 2013.
Qinglin Guo and Ming Zhang, question answering system based on ontology and semantic web, G. Wang et al. (Eds.): RSKT 2008.
Bassam Hammo, Hani Abu-Salem, Steven Lytinen, QARAB:A question answering system to support Arabic language, DePaul university, 2002.
Fatma Zohra Belkredim, Ali El Sebai, An ontology based formalism for the Arabic language using verbs and their derivatives, communications of the IBIMA, volume 11, 2009.