Core Concepts
A novel context-aware approach is proposed to recommend next services in a workflow development process, through learning service representation and service selection decision making behaviors from workflow provenance.
Abstract
The paper presents a novel context-aware approach to recommending next services in a workflow development process. The key highlights are:
The process of composing a scientific workflow is formalized as a step-wise procedure within the context of the goal of the workflow. The problem of next service recommendation is mapped to next word prediction in natural language processing.
Historical service dependencies are extracted from scientific workflow provenance to build a knowledge graph. Service sequences are then generated based on diverse composition path generation strategies (intra-workflow and inter-workflow).
The generated corpus of composition paths are leveraged to study previous decision making strategies. A goal-oriented context-aware LSTM (gLSTM) model is developed to predict the probabilities of services to be selected at the next step, considering both the sequential context of selected services and the goal requirement of the workflow.
The trained gLSTM model is used to recommend top K candidate services during the workflow composition process. Extensive experiments on a real-world repository have demonstrated the effectiveness of this approach.