Raymond Bentinck

Semantic Web provides a benchmark by which a search application can make an (ostensibly) objectively formed judgment, based on the criteria laid down in a search ontology. Very simply, an ontology is a logical structure that enshrines categories, or hierarchies. Search based applications are an increasingly familiar feature of the Semantic Web, and are the accepted means by which values, categories and conceptual relationships with other objects are assigned to search objects, using ontologies.

The Semantic Web provides the model to extract ‘value‘ data from the context of search objects and to classify relationships from text ‘in the wild’, as easily as from databases. So, if a consumer expresses a certain level of emotional content, it’s possible for a search-based application to ‘score’ information and present that as meaningful, organised content. The knowledge contained in the ontology can be linked into real time web expression and trends, derived from, perhaps, the past 15 minutes or two days, which can then be dynamically incorporated into the ontological view of the organisation.

A Cultural Problem

Increasingly, however, humans are unable to deal with the information overload with which we are presented online. If a consumer wants to find a hotel in London, searching for hotels in London or Knightsbridge will throw up data from addresses, chain websites and, perhaps, the addresses of people who’ve commented online. Beyond that, search, as it’s commonly used, is little use. If someone wants a good value, informal hotel, search results will likely include The Dorchester Grill – informal, yes, and quite possibly good value, but not likely to be thought of as such by the average person’s judgement.

Consumers can no longer trawl through endless aggregators and guides and make convoluted Boolean searches. We want – and need – to be able to incorporate subjective information and to take into account cultural factors and value judgments made in relation to different aspects of an item of interest.

Semantic Potential

Particularly fascinating are the implications of the combined capabilities of Semantic Web search applications and the technology to extract and apply value judgments to the questions that web users and customer want to address.

Consider this: the Semantic Web is inextricably linked with social media because of its ability to categorise, score and forge relationships between new data in real time. As people now make spur-of-the-moment comments and post judgments online in social media, the ability to make effective use of those judgments, to aggregate and to incorporate new subjective input into ontologies in real time, becomes increasingly important, whether we’re simply aggregating responses to a business, event or product, or managing relationships with business prospects or existing customers.

Introducing the Subjective Web

First generation semantic tools are already incredibly powerful. Take, for example, Real Travel’s presentation of practical tips, geographical data and travellers’ responses extracted from travel blogs, to people planning itineraries and trips abroad. Highly relevant information is presented in a completely usable format by a very sophisticated ontology. It is possible to virtually always extract the tone or responses of visitors to a location, business or place, and be confident that the blog will have the relevant locations automatically extracted, itineraries extracted on the basis of text and advice, or warnings presented to anyone thinking of following in someone’s footsteps.

The emergent Subjective Web takes this one step further. An early example of this development is Yellow Pages Group’s Urbanizer, a restaurant recommendation widget for the iPhone. Its ability to generate meaningful restaurant recommendations was not just based on the Francophone background of its developers. Its capability to discern how formal (or otherwise) a restaurant is, or, for example, how trendy a restaurant is perceived to be, is evident from the fact that it doesn’t just put restaurants in boxes, but operates on the basis of a sliding scale interface; formal and informal restaurants aren’t presented as if in separate silos in an ontology. They are scored and cross-referenced with a search objects’ ‘scores’ in other categories, such as perceived value, speed of service or place on a sliding scale between homely or classical cuisine. Whilst this is just one, very early, example of an application enshrining crowd sourced judgement and subjective input, it provides an indicator of the future direction of the Semantic Web and our cultural expectations of the Internet itself.

About the author

Raymond Bentinck is UK and Ireland Managing Director at Exalead.