Web-based information problem-solving has been recognised as a critical ability for learners. However, the development of students' abilities in this area often faces several challenges, such as difficulty in building well-organised knowledge structures to support complex problems that require higher-order skills (e.g., system thinking). To resolve these issues, this study employs a semi-automatic tool that supports query expansion-based concept mapping (QECM) for assisting learners' web-based information problem-solving. The query expansion technique aims to recommend relevant concepts and linking words for building the map. The linking of concepts also uses non-taxonomic relationships for visualising a systemic model to develop complex problem-solving. An experiment was conducted by randomly dividing 50 participants into two groups, QECM (experimental) and conventional keyword-based search system, (control), to compare their performance during web-based information problem-solving tasks. The results show that the QECM system facilitated participants in extending their queries so as to enhance the comprehensiveness of their constructed concept maps. The QECM also improved the participants' information problem-solving performance by bridging concepts of an assigned task. The findings imply that learners using the QECM system can focus on the higher-order tasks of problem-solving and be better engaged in exploring real-life problems with the web.