Online support centers are emerging as a cost-effective and innovative solution designed to enable end-users to resolve technical problems more effectively without relying on live support from contact center agents. However, the capacity limitation of corporate knowledge bases prevents online support centers from effectively resolving user problems. In addition, traditional textual search techniques employed by most online support centers fall short from accurately interpreting user queries due to the ambiguity of user requests and the heterogeneity of technical problems. In this paper, we present SolutionFinder, an autonomous framework, which dynamically integrates online resources to enrich the knowledge base for IT support systems. SolutionFinder provides context-aware search support to remove the textual ambiguity embedded in user queries. Furthermore, SolutionFinder transforms solution documents into solution paths to analyze their similarity to provide high-quality solution recommendations. Evaluation results suggest by leveraging our proposed algorithms, a support service can accurately locate Web solution resources and provide high-quality services.