The software quality control is one of the main topics in the Software Engineering area. To put the effort in the quality control during the specification phase leads us to detect possible mistakes in an early steps and, easily, to correct them before the design and implementation steps start. In this framework the goal of SAREL system, a knowledge-based system, is twofold. On one hand, to help software engineers in the creation of quality Software Requirements Specifications. On the other hand, to analyze the correspondence between two different conceptual representations associated with two different Software Requirements Specification documents. For the first goal, a set of NLP and Knowledge management tools is applied to obtain a conceptual representation that can be validated and managed by the software engineer. For the second goal we have established some correspondence measures in order to get a comparison between two conceptual representations. This information will be useful during the interaction.