Applying web intelligence to improve corporate websites
Case study Knowledge Transfer Partnership
In a KTP with the University of Westminster, ActiveStandards has improved quality assurance for corporate web contents, by incorporating state-of-the art semantic technologies and information extraction techniques into their products and services. They are now able to expand their leading position and deliver advanced capabilities for corporate web quality management.
ActiveStandards (formerly Magus) is the leading provider of managed solutions for optimising website quality and compliance, used by large corporations such as Shell, Unilever, Rolls-Royce and ING. ActiveStandards’ website quality monitoring solutions help companies track and manage compliance with corporate quality standards. These solutions address quality assurance for corporate web contents – they help clients analyse large web estates of web contents to gain insight that will enable them to target the most critical quality, compliance and optimisation issues.
The project aimed to enhance the existing ActiveStandards platform by introducing an in-house semantic tier and text-mining techniques for analysis of web contents. This meaning-making technology would significantly improve ActiveStandards’ products and solutions by offering new ways of interpreting and analysing information on the web.
The semantic tier, through its text-mining techniques, would deliver new functionality in the current product range and also create the opportunity to develop or enhance new products incorporating this aspect of web intelligence.
Embedding knowledge from emerging concepts in semantic technologies would increase competitive advantage, aid retention of clients, and enhance platform capabilities in preparation of a future, more semantics-driven web.
How did we help?
We applied web intelligence techniques, such as text-mining and ontology learning, to detect inconsistencies and erroneous information in corporate web contents.
Significant achievements of the KTP include:
- We successfully developed an in-house semantic tier that provides powerful semantic analysis and annotation (text analysis, named entity recognition and disambiguation, relation extraction).
- The tier integrates into the common Magus platform.
- We delivered quality improvements across multi-domain content using a variety of approaches (linked data enrichment of knowledge bases, refinement of JAPE rules, improvement POS components using machine-learning based module).
- We developed operational tooling for maintenance and control of generated semantic data.
- The semantic tier features surfaced in Magus main quality and compliance product.
- We delivered new content and SEO insight features for ActiveStandards.
- We removed the dependency on third-party Calais services.
- We developed benchmarking methods – 'gold standard' corpus and automated test packs that provide quality assurance.
- We improved performance, scalability and stability, significantly increasing the throughput of semantic services.
The business results
As a result of the KTP project, ActiveStandards has been able to add an entirely new semantic tier to its shared in-house platform, which provides functionality to semantically analyse cross-domain unstructured content. ActiveStandards has also been able to extend their product portfolio to customers by providing deeper analysis of content on their web estates.
While the core business of the company has not changed its focus on web quality management and compliance solutions, as a direct result of the KTP project the in-house semantic capability we developed over the last 24 months has extended the product offering in an area that sees rapidly growing customer awareness and commercial importance.
In its final evaluation, the Technology Strategy Board graded the KTP with grade B (very good), the second highest in a five-grade scale.
Read more about the KTP on ActiveStandards' web governance blog.
We've laid a great foundation with the work done via the KTP, which we aim to build upon. Simon Lande, CEO
Who was involved
Read more about who was involved with this project:
Take a look at our case studies based on the research theme of your choice.