Multi-Context Incremental Reasoning over Data Streams

Data streams are generally generated on-the-fly and contain valuable knowledge to support the decision process in many use cases, such as smart cities or traffic monitoring. To this end, reasoning over streaming data suggests a reconsideration of the concept of a solution to a problem, and how to obtain it. As data in streams change continuously, all the conclusions based on expired data must be retracted, and additional or new information might also arrive and output additional derivations. Multiple works have investigated the use of Answer Set Programming (ASP) for streaming data. However, they generally do not provide window mechanisms, which are key factors in stream processing. Ticker is one of the recent solutions provided to tackle this issue. It is based on LARS, a Logic-based Framework for Analytic Reasoning over Streams, and has two reasoning strategies. One relies on Clingo for repeated solving, and the other uses a Truth Maintenance System to perform model updates. Despite the significant achievements that have been made, there is still room for improvement, especially when considering real-life applications. In this paper, we propose a new stream reasoning engine based on Ticker that overcomes its limitations. The proposed framework has been implemented and evaluated with a real-world benchmark, and a use-case scenario implementation shows promising results.