This paper presents a systematic data-driven approach to assisting situational application development. We first propose a technique to extract useful information from multiple sources to abstract service capabilities with set tags. This supports intuitive expression of user’s desired composition goals by simple queries, without having to know underlying technical details. A planning technique then exploits composition solutions which can constitute the desired goals, even with some potential new interesting composition opportunities. A browser-based tool facilitates visual and iterative refinement of composition solutions, to finally come up with the satisfying outputs. A series of experiments demonstrate the efficiency and effectiveness of our approach. Data-driven composition technique for situational web applications by using tag-based semantics in to illustrate the overall life-cycle of our “compose as-you-search” composition approach, to propose the clustering technique for deriving tag-based composition semantics, and to evaluate the composition planning effectiveness, respectively.
Compared with previous work, this paper is significantly updated by introducing a semi-supervised technique for clustering hierarchical tag based semantics from service documentations and human-annotated annotations. The derived semantics link service capabilities and developers’ processing goals, so that the composition is processed by planning the “Tag Hyper Links” from initial query to the goals. The planning algorithm is also further evaluated in terms of recommendation quality, performance, and scalability over data sets from real-world service repositories. Results show that our approach reaches satisfying precision and high-quality composition recommendations. We also demonstrate that our approach can accommodate even larger size of services than real world repositories so as to promise performance. Besides, more details of our interactive development prototyping are presented. We particularly demonstrate how the composition UI can help developers intuitively compose situational applications, and iteratively refine their goals until requirements are finally satisfied. Data-Driven Composition for Service-Oriented Situational Web Applications
In our previous work we have designed a technique to extract tags by mining service specifications (including WSDL, Web API documents and web pages that contain references to web services) and collecting human-generated contents (including comments and queries). Several web services tagging approaches have been proposed, for example the FCA tagging system most of them annotate web services manually. Manual tagging is a time consuming work. Moreover, several existing systems can recommend tags for web services based on existing handmade tags such as the approach these systems consider nothing about similarities between tags and web services. Another problem in these systems is that if there is no handmade tag, they cannot work at all. Another system CDKH in can generate tags for web services automatically, but the system doesn’t use existing handmade tags of web services.
These different ways are combined in tagging tools that the tag-based platform facilitates. Moreover, inside of the platform and due to the preferences of the users, different tagging behaviors exist that actually obstruct the automated interoperability among tag sets. Despite the fact that the systems offer solutions to aid the understanding of the folksiness that the users collectively build (tag clouds, tools based on related tag ideas, collective intelligence methods, data mining, etc.) Although tagging shows potential benefits, personal organization of information leads to implicit logical conditions that often differ from the global one. Tagging provides a sort of weak organization of the information, very useful, but mediated by the user’s behavior. Therefore, it is also possible that user’s tags associated with an object do not agree with the other users tags.
We propose a heuristic graph-based planning algorithm within polynomial-time complexity. When the developer selects a tag from the tag cloud or input a keyword as the initial query request qi , the planning algorithm first computes the cost of achieving each tag starting from qi by conducting a forward search. Such a Depth-First Search step constructs all possible Tag Links that can perform the final goal. Based on the results above, the planning algorithm then approximates the sequence of Tag Links that connects qi to the final goal by a regression search step the tourist takes geographical locations of hotel, restaurant, bars and museum, we cannot give the reasonable order for visiting these places. Preferences, quality, ordering and other constraints would be helpful to improve the plan quality and performance. Due to the popularity and simplicity of tags, our tag-based service model can be extended, where all these constraints can be also presented as tags.
Our approach relies on the popularity of tags on the web. The primitive of tag-based composition of flow applications was first proposed in the MARIO system. Tag-based search is a hot topic in the research body of information retrieval and data mining. Most of existing research works focus on processing tags from popular social networking sites like Del.icio.us, Twitter and flickr. To best of our knowledge, few works have been made in the area of existing service-based applications. The primitive of tag-based composition of flow applications was first proposed in the MARIO system. Some recent works try to leverage tag-based service discovery, but not fully consider the hierarchy relationships of tags.
Our approach provides a systematic way for extracting useful tags from service documents and user generated annotations, by fully considering the unique features of web services like interface naming rules and developer preferences. Besides traditional similarity-based measurement, the clustering process is also controlled by the probability of tag occurrence and its own property, without any needs of training data. It should be noted that, we currently make simple mapping between our top-level tags to WordNet. However, the search results seem to be satisfying in regular cases.