Wood and Sommerville [13] propose a frame-based software component catalogue, attempting to capture the meaning of software components. The catalogue consists of a frame for the basic function that a software component performs, with slots for the objects manipulated by the component. The frames are not used to parse input sentences. Instead, the retrieval system presents the frames as templates to be filled in by the user via menus. The time- consuming search through the possible fillers is left to the user. The indexing task, i.e. the construction of frames for the software component catalogue is performed manually.
Relatively few software reuse systems take advantage of the conceptual information available in the natural language documentation of software components to improve retrieval effectiveness. Existing systems may be classified in two basic groups: free-text indexing reuse systems, and knowledge-based reuse systems.
Free-text indexing systems automatically extract keyword attributes from the natural language specifications provided by the user and these attributes are used to locate software components. Similarly, software components are classified in the software library by indexing them according to keyword attributes extracted from the natural language documentation of the software components. These systems work at the lexical level, so they ignore much of the useful syntactic and semantic information available in a description in natural language. No semantic knowledge is used and no interpretation of the document is given. The GURU system [6] follows this approach.
Knowledge-based reuse systems make some kind of
syntactic and semantic analysis of natural language
specifications without pretending to do a complete
understanding of documents. They are based upon a
knowledge base which stores semantic information about
the application domain and about natural language itself.
These systems appear more powerful than traditional
keyword retrieval systems. However, they usually require
enormous human resources: knowledge bases are created
for each application domain and populated from scratch.
The LaSSIE [1] and NLH/E [11] systems follow this
approach.
8 Final remarks
Site Hosting: Bronco