Developing an Efficient Mechanism for Retrieving Documents Using Automatically Generated Metadata

Abstract

Efficient generation of metadata and retrieval is becoming highly challenging with the increase in volume of resources. Huge recall with low precision is the problem of most document retrieval systems. Effective automatic generation of metadata can reduce time required in creating metadata manually. This paper attempts to develop asearch algorithm to retrieve documents using automatically generated metadata. The fifteen elements of Dublin Core Metadata are discussed and addition of another element consisting of popular search keywords for retrieving the document has been suggested. The keywords in the suggested extended metadata element can be grouped according to geographic location for personalizing the search results. Precision increment and recall decrement can be achieved by retrieving books of specific topic-ids for which maximum words match with the searched keywords. Different stemming mechanisms for the search keywords have also been discussed. A flowchart of the proposed algorithm is given at the end.