We have been involved in a number of large-scale software development projects, which might lead to loss of millions of dollars if failed. The quality of SRS (Software Requirements Specification) is the key to success of the software development. Review and inspection are common practices for the verification and validation of SRS. However, verification techniques used in projects might be characterized as ad hoc. In this article, we propose requirements clinic, a third party inspection methodology for improving the quality of the SRS. In order to systematically inspect a SRS, we developed a perspective-based inspection methodology based on PQM (Pragmatic Quality Model) of SRS. PQM is derived from IEEE Std. 830 from the perspective of pragmatic quality. To inspect a SRS according to PQM, we identified 198 inspection points, which lead to a quality score between 0 and 100. The requirements clinic advises to the requirements engineering team by a comprehensive quality inspection report including quality score, benchmark and SRS patterns for improvement. Since 2010, we have been practicing the methodology to a variety of development projects, and revealed an average of 10.6 ROI in 12 projects. We also discuss the feasibility of the methodology and lessons learned from the practices.

The aim of our research is to empirically analyze the impact of requirements trace ability quality to the productivity of the enterprise applications development. We analyze trace ability links of Requirements Specification Documents (RSDs) collected from the enterprise applications development. These RSDs are classified into two groups in terms of project performance. One group includes RSDs of significant cost overrun projects, another group includes RSDs of normal range cost projects. Our analysis revealed that high quality of trace ability in RSDs significantly reduces the cost overrun in the enterprise applications development.

Information extraction is the task of automaticallypicking up information of interest from an unconstrained text. Informationof interest is usually extracted in two steps. First, sentence level processing locates relevant pieces of information scatteredthroughout the text; second, discourse processing merges coreferential information to generate the output. In the first step, pieces of information are locally identified without recognizing any relationships among them. A key word search or simple patternsearch can achieve this purpose. The second step requires deeperknowledge in order to understand relationships among separately identified pieces of information. Previous information extraction systems focused on the first step, partly because they were not required to link up each piece of information with other pieces. To link the extracted pieces of information and map them onto a structuredoutput format, complex discourse processing is essential. This paperreports on a Japanese information extraction system that merges information using a pattern matcher and discourse processor. Evaluationresults show a high level of system performance which approaches human performance.