Abstract. Natural Language Processing techniques for text-mining and information retrieval are finding application in the analysis of many kinds of documentation, from technical documentation to World Wide Web. Particularly, Functional Analysis techniques are based on the extraction of the interactions between the entities described in the document: these interactions are expressed as Subject-Action-Object (SAO) triples (obtainable using a suitable syntactic parser) which represent a concept in its most synthesizing form. In this work, the techniques developed for a functional analysis of patents and their implementation in the PAT-Analyzer tool are presented. The same technique has been properly tailored and applied to the analysis of software requirements documents. Current work in the direction of the development of a SAO-based Content Analysis of technical documentation is presented.
The experience reported in this paper relates to an evaluation of the automatic generation of C code from the Specification and Description Language (SDL) specification of embedded applications. The evaluation has been carried out by comparing the automatically generated code with the manually implemented code, both compliant to the same SDL specification: this comparison is based on a selection of metrics measured on both codes by means of commercial static analysis tools. Notwithstanding the different structure of the two codes, we appropriately selected and aggregated the obtained results in order to use them as indicators of code size, control flow complexity and integration flow complexity. For a better comparison of the two codes, we have also introduced a novel complexity metric, which compares the control flow complexity with the integration flow of the two different software architectures. The aim of the paper is not merely to evaluate the code generator used, but rather to propose a set of techniques that can be used to conduct similar evaluations.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.