UcEF for Semantic IR: An Integrated Context-Based Web Analytics Method

Introduction

Information Retrieval (IR) systems demonstrate the IR paradigm that allows users to search and retrieve relevant documents that contain the information they need. The Web Search Engine (WeSE) is the most visible example of IR systems since the majority of Internet users rely on it to search the Web (Akhigbe, 2015). Search on the Web is typically modelled as tasks-centric experiences considering the dynamic context of the Web. This fluxing context informs how IR systems operate, and how they are investigated and used (Akhigbe, 2015). This tasks-centric - usage - experiences are bound to change as the inclusive Internet (i.e. Internet of Things - IoT) - that is touted as able to offer limitless possibilities regarding connectivity (Malik and Malik, 2020) - enters the stage. More dynamic system collaborations, and cooperation in the form of integration and interoperability (Akhigbe et al., 2016), and diverse IoT applications with varied contexts of deployment with a frequency of continuous change and Disparate Information Sources (DIS) will be prevalent. Enormous amounts of data in dissimilar and badly expressed format will be delivered by the DIS, which will be difficult to understand and exploited by devices and systems. More concerns such as standardization, discovery, and poor IoT information resource descriptions are also expected as causes of retrieval difficulty (Rhayem et al., 2020).

The solution to the foregoing is the Semantic Web IoT (SW-IoT) technologies, which are fast emerging. However, the Semantic Web (SW) is still wanting of a common information structure and standard to make the extraction of meaningful information possible. The keyword-based structure of the SW makes it semantically incapable to resolve issues of context (Malik and Malik, 2020). Its lexicon and semantic structure, therefore, need reworking to shore up its potential to deal with Context and Meaning (C&M) to avoid document misinterpretation. More complex and evolving incompatible (internal) semantics than what exists currently on the Web from varied sources of information will also be avoided. This highlights the need for a cognitive-based approach to the semantic description of terms to aid the retrieval of ambiguous terms based on “relevance”. Unfortunately, “relevance” is still considered based on the frequency of query terms expressed in statistical measures (Lashkari et al., 2018). This requires a technical response that approaches semantic retrieval differently from the traditional IR topic and key-words based. The exercise of evaluation has contributed greatly to this regarding retrieval speed, particularly from the non-cognitive context. This is because, in the IR domain evaluation is important for system tuning and improvement (Ferro and Silvello, 2018). However, both the non-cognitive and Cognitive Context of IR Evaluation (CCoIRE) needs rethinking in terms of methodology considering the emerging dynamics to be introduced by the concepts of the SW and the IoT about search and retrieval. The focus of this chapter is on the CCoIRE, which could be exploited to complement the technical response to enhance the potentials of Semantic Information Retrieval (SIR) on the SW. For example, one of the widely used approaches to evaluate SIR is the Information Content Approach (ICA) (Jiang, 2020). However, despite its use, a clear solution is still elusive since the approach is topically inclined. This highlights the need for a “Contextual IR Approach” (CoIRA) within the CCoIRE that promotes the association of Users Information Need (UIN) with semantic concepts to represent real users (Butavicius et al., 2019).