Ontology revisited

Employing Hyper-graph or Graph to build the Knowledge Net:

Hypergraph (Formal Definition): A hypergraph G can be defined as a pair (V, E), where V is a set of vertices, and E is a set of hyper-edges between the vertices. Each hyperedge is a set of vertices: E ⊆ {{u, v, ...} ∈ 2V}. (Hyperedges are undirected.)

A Hyper Graph database is a generalized graph of entities. The generalization is two-fold:

• Links/edges "point to" an arbitrary number of elements instead of just two as in regular graphs.
• Links can be pointed to by other links as well.

• A graph can cater the basic ontological problems (e.g. Storing Synonyms, Super-concepts – sub concepts and other basic relations). Using graph instead of traditional relational database to implement the Knowledge Net can provide solutions for the following problems:

1. Knowledge Net need not be strictly predefined. It can be flexible enough to enhance or    augment later.
2. Previously extracted Information or processed Text need not be suffered (re-processed) by further enhancement and/or augmentation of Knowledge Net.
3. Normalization may be reduced enough and hence saving processing time.

• But a Hypergraph can be utilized to cater more realistic approach in the same area. E.g:

Case 1: If the input Query is something related to Terrorism the system is supposed to search for all the terrorism related incidents including bomb blasts from the processed documents.

In the current system to cater such problem blast have to be declared as a sub-concept or synonym of terrorism though blast is not a terrorism in all cases. So by doing this “blast of a gas cylinder” will also be qualified as terrorism will be displayed as a relevant result which is not correct. Instead if the knowledge is stored like the following then the two different types of blast can be distinguished and results can be filtered properly.

• Storing Processed Paragraph as an Object or as an equivalent XML Structure in the Database:

1. A Collection of ISentence (as shown in the following Diagram) Objects can be constructed from the finally processed Paragraphs.

Case 1: If the database of use is SQL Server 2008 or like, then it can be stored directly in the Database.
 

Case 2: Alternatively, since the ISentence Objects can be converted into XML and Vice Versa, the paragraph can be stored as XML String into the Data Base. At the time of retrieval the objects have to be reconstructed again to facilitate the retrieval.

Class Diagram for Sentence Representation

[* DateTimeOfOccurance will be another Property in the IsemanticAction interface]

• Summarized Benefit:

1. Better Knowledge Abstraction approach (Flexible, Upgradable and Optimized).
2. Faster Retrieval 
3. No need for further reprocessing of extracted information in case of Knowledge Update. 
4. More Appropriate Retrieval and Relevance ranking. 
5. Knowledge and information representation in terms of Hypergraph or Graph leads to the road to NLG (Natural language generation) and hence can lead to develop more appropriate Summarizer, Question-Answering System etc.