Justified Assessments of Service Provider Reputation

Justified Assessments of Service Provider Reputation (JASPR) aims to improve the way that services are discovered, selected and used by providing rich, personalised reputation assessments of services with the rationale behind those assessments. It is particularly targeted at giving small and medium-sized enterprises (SMEs) better exposure to large clients by reducing clients' reliance on extensive market histories or opaque online reviews that do not account for personalised needs. While large companies can rely on brand influence to bring clients to their services, it is difficult for SMEs to gain market access, especially newer businesses with little history for clients to draw on. This can mean that an SME is overlooked even when providing a service directly matching the client's needs. From the customer side, the project will allow more intelligent service procurement, based on rich reputation assessments reflecting the actual performance of providers, with less bias from branding and superficial reviews.

More generally, in any service-based system, an accurate assessment of reputation is essential for selecting between alternative providers. Existing methods typically assess reputation on a combination of direct experiences by the client being provided with a service and third party recommendations, with the reputation expressed as a numerical score or probability estimate. They do not allow the opportunity to interrogate an assessment to find out why a particular assessment is made, and so whether it is appropriate to a new service selection requirement, and they exclude from consideration a wealth of information about the context of providers' previous actions that could give useful information to a customer in selecting a service provider. For example, there may be mitigating circumstances for past failures, or a provider may have changed their organisational affiliation. These limitations are of particular significance in marketplaces involving both newer and more established service providers. New providers are often disadvantaged in a marketplace since a single negative review can disproportionately harm their reputation, and customers are unable to accurately assess the risk associated with new providers compared to those that are established. To make richer reputation assessments that take into account the context of past service provisions, this context must be modelled and recorded, and can be described as the provenance of the provision.

In the proposed project, we will use provenance records as a source of information on which a more nuanced reputation mechanism can be based. We will define the supporting algorithms and software infrastructure to allow this rich reputation information to be captured, analysed and presented to clients.

Decisions on which providers to use for particular services, such as suppliers of goods or utilities, are often based on reputation. This favours long-established providers with recognised brands, regardless of whether they provide the most appropriate or best value service. Reputation is often based on past customer reviews or recommendations which, while valuable, has important failings. For example, in the online world every review is typically treated equally (by the review site) regardless of its relevance to the particular need. Also, the aspects of a service influencing a past customer's rating may not be the aspects that are most important to the potential customer.

Our aim is to develop the algorithms and technology that allows decisions on service providers to be based on rich, personalised, relevant reputation information, where less information from past customers is required for reputation assessments because more useful and objective information is derived from each service provision. There are many areas in which the results of the proposed research can be applied, and this is expected to create opportunities for establishing new collaborations with business and industry. The methods developed for provenance-driven reputation can be applied in any situation where services must be selected, and can benefit stakeholders across the system, including: (i) better exposing smaller and newer businesses to potential custom, (ii) enabling organisations needing to make use of a set of providers, drawn from a wide pool of possibilities, to obtain better services tailored to customers' needs, (iii) providing algorithms and technology that can be used by service middleware providers in developing service management infrastructure, and (iv) offering end consumers better and more informed choice of services. Given the widespread adoption of service-oriented systems, these potential applications of the research indicate the strategic importance of the project. There is potential for IP generation, and the appropriate exploitation of IP will be considered as part of the project management.

The scientific and technology outputs of the project are generally applicable, and the involvement of Black Pepper Software (BP) will not only enable us to effectively evaluate the outputs, but will also provide technology impact in the areas of service middleware in general and for the logistics domain specifically. As a software company with extensive experience of service-oriented systems in a range of applications, BP's involvement will promote the adoption of the developed technologies and will expose the project results to industry, both in its own terms and as an exemplar of the value and benefits to be gained. BP are regular contributors to industry conferences and exhibitions and are well placed to support dissemination to industry. Through the case studies, supported by BP, the project will also make specific technology contributions that are applicable to the logistics domain.

We aim to build new links that can lead to both potential application of the outputs through adoption by industry and to future research proposals with strong industry involvement. A reputation workshop will be organised to which relevant companies, who are the potential users of the research results, will be invited. In the medium term, it is envisaged that this project may result in, for example, technology transfer through new Knowledge Transfer Partnerships.