Recovering requirements traceability is a resource and time consuming process that is difficult to automate. This work investigates the application of information retrieval algorithms for automated detection of traceability links between requirements. It will be assumed that the recognition rate will depend on the type of the traceability link. The author investigates the following types of traceability links:
structured dependencies, constraint dependencies, and cost/value dependencies. To evaluate the hypothesis a vector space model and latent semantic indexing algorithm are analyzing two case studies with template based requirements. The result shows that structured dependencies are best recognized, despite a generally low detection rate.