Semantic Search

Created on March 13, 2013, 8:58 a.m. by Hevok & updated by Hevok on May 2, 2013, 5:14 p.m.

Semantic Technologies are able to support Search in the Web and also enable completely new ways to Search like for example Exploratory Search.

Semantic Search means to improve traditional Web-based keyword based Search Information Retrieval with the Help of Semantic Technologies. In Keyword-based Search one is most times dealing with Symbols and Natural Language. In Natural Language the Term stands for an Object in the real World, but this Terms can be ambiguous and it depends on the Meaning and the Mapping to the right Concepts, right Ontology, right Knowledge Representation whether Sender and Receiver in the Communication are able to communicate and are successful and they have to refer to the same Concepts which is quite important as it makes a Difference.

Whenever we are talking about a Term, we are referring to an Entity and Knowledge related to this Entity. If we are talking about a Term we have the URI of an Information Resource in the World Wide Web that gives us Metadata, i.e. Information about the Object we are talking about. This Concept can be applied in Information Retrieval.

Traditional Search retrieves a lot of Results there are some positive Matches among them but also many false positives i.e. about similar named Concept or even just about Concepts that are somehow related to a similar named Concept. This is the way Google Searches for Multimedia or Images.

One has to take into account that we are able to apply Semantic Web Technologies for Search Technologies. So one has to define what does Semantic Search really mean. Semantic Search has lots of Definitions. For our purpose we refer to Semantic Search when we have Text-based Metadata / Annotations and they are annotated with Semantic Entities which means we have for example Text, Pictures or we have Videos and they are annotated with Metadata and this Metadata refers to Semantic Entities which have a connection to an Ontology which is a Knowledge Representation and tells one about the Meaning of these Things. So one is not anymore talking about Keyword-based Retrieval, because we do not look for Things or Mappings / Matches within Strings of the Language, we are looking for Entity based Information Retrieval which means in our Query we phrase our Queries in terms of Entities and the Answer will also be Documents that are annotated with Entities and when these Entities are found within the Documents we will give back these Documents that we have found. What we further can do is to make use of Semantic Relationships for Example, so Content-based Similarities can make Sense as one can also give back not only exactly matching Results but Results that are somehow related or somehow similar to the Query. Then, this Metadata via Semantic Annotation is of course interoperable and can be used for Content-based Description and also for structural / technical Descriptions.

So the overall Goal of Semantic Search in our Purpose is to improve the Quantity and Quality of Information Retrieval which means more complete and more precise Results. We want to increase first the Recall and we want to increase the Precision and we do this by Mapping our Information Resources against Semantic Entities and by applying Semantic Search, which means Entity-based Search on our Documents, the Semantic Annotated Documents. This results into an Improvement of Recall and Precision and this would be some kind of Semantically Enhanced Classical Information Retrieval.

Overall Goal: Quantitative and qualitative Improvement of Information Retrieval.

Semantic Technologies fit in many points of Information Retrieval for their Improvement. First of all we have the Possibility to extend or refine additionally the Query String, i.e. to enable more precise and complete Search Results. Secondly one has also the possibility to determine Cross References, so to complement the Search Results with additional or similar Information which would be useful for the User. Moreover one can enable Exploratory Search which means one can show new Paths or new ways within our huge Repository of Information and Knowledge that we have. For example in the Web there are so many Documents it is really difficult to get an Overview or it is really difficult to navigate between these Documents. Of course one has Links, but if one follows Links they are not Topic-based, one has not the Possibility to follow a Link according to a Topic. This would be something that Exploratory Search enables. There one would have means of Visualization and Navigation of the entire Search Space. Last, one can also apply Reasoning, so maybe some of the Results that are possible and are Relevant are not really obvious and one has to think about and really enable Reasoning and Entailment to infer or deduce the Results that are based on implicitly given Information. This would be really the ultimate Thing to do which is also the most complicated to enable Reasoning in order to complement the Search Process, but this is also of importance.

Semantic Metadata enable Improvement of traditional Keyword-based Retrieval by

  1. Query String Extension/Refinement Enables more precise or more complete Search Results
  2. Cross Referencing Enables to complement Search Results with additional associated or similar Information
  3. Exploratory Search Enables Visualization and Navigation of the Search Space
  4. Reasoning Enables to complement Search Results with implicitly given Information
semantic-search.jpg

Tags: web, information, searching, entity, semantics, retrieval
Categories: Concept
Parent: Search
Children: Cross Referencing, Query String Extension, Query String Refinement, Reasoning

Update entry (Admin) | See changes

Comment on This Data Unit