As an effective tool to alleviate information overload on the service platform, sequential recommender systems aim to predict the service in which users are interested by analyzing their historical behaviors. To leverage transition patterns among services, some solutions apply the graph attention networks for service representation learning via neighborhood information aggregation. However, existing solutions struggle to sufficiently leverage the graph structure due to two significant challenges. Firstly, some high-correlation services may not appear in adjacent positions and thus have no connections, which makes it difficult to aggregate comprehensive information in graph learning. Secondly, the attention network parameters are randomly initialized, which makes the information propagation unstable in the early training phase. To tackle the two challenges, we propose a novel neighborhood-augmented graph collaborative attention network (NA-GCAN). For the former challenge, we augment the graph structure by screening potential neighbors with high correlation for each service node based on the attention network, to ensure effective aggregation of global-wise information. For the latter challenge, we exploit the co-occurrence information to pre-train service embeddings and the transition information to guide the information propagation. In addition, we devise a novel two-stage learning strategy to enable a warm start for model training and make full use of the augmented graph structure. Extensive experiments have demonstrated the superiority of our proposed NA-GCAN.

Due to the exploding growth in the number of web services, mashup has emerged as a service composition technique to reuse existing services and create new applications with the least amount of effort. Service recommendation is essential to facilitate mashup developers locating desired component services among a large collection of candidates. However, the majority of existing methods utilize service profiles for content matching, not mashup descriptions. This makes them suffer from vocabulary gap and cold-start problem when recommending components for new mashups. In this paper, we propose a two-step approach to generate high-quality service representation from mashup descriptions. The first step employs a linear discriminant function to assign each term with a component service such that a coarse-grained service representation can be derived. In the second step, a novel probabilistic topic model is proposed to extract relevant terms from coarse-grained service representation. Finally, a score function is designed based on the final high-quality representation to determine recommendations. Experiments on a data set from ProgrammableWeb.com show that the proposed model significantly outperforms state-of-the-art methods.

An increasing number of web services are being invoked by users to create user applications (e.g., mashups). However, over time, a few good services in service systems have become deprecated, i.e., the service is initially available and is invoked by service users, but later becomes unavailable. Therefore, the prediction of service deprecation has become a key issue in creating reliable long-term user applications. Most existing research has overlooked service deprecation in service systems and has failed to consider long-term service reliability when making service recommendations. In this paper, we propose a method for predicting service deprecation, which comprises two components: Service Comprehensive Feature Modeling (SCFM) for extracting service features relevant to service deprecation and Deprecated Service Prediction (DSP) for building a service deprecation prediction model. Our experimental results on a real-world dataset demonstrate that our method yields an improved Area-Under-the-Curve (AUC) value over existing methods and thus has better accuracy in service deprecation prediction.

Semantic extraction is essential for semantic interoperability in multi-enterprise business collaboration environments. Although many studies on semantic extraction have been carried out, few have focused on how to precisely and effectively extract semantics from multiple heterogeneous data schemas. This paper presents a semi-automatic semantic extraction method based on a neutral representation format (NRF) for acquiring semantics from heterogeneous data schemas. As a unified syntax-independent model, NRF removes all the contingencies of heterogeneous data schemas from the original data environment. Conceptual extraction and keyword extraction are used to acquire the semantics from the NRF. Conceptual extraction entails constructing a conceptual model, while keyword extraction seeks to obtain the metadata. An industrial case is given to validate the approach. This method has good extensibility and flexibility. The results show that the method provides simple, accurate, and effective semantic interoperability in multi-enterprise business collaboration environments.