Real-time access to complex knowledge is a business driver in the contact centre environment. In this paper we outline for the domain of telecom technical product support a knowledge sharing paradigm in which a desktop client annotates named entities in technical documents with canonical names, class names or relevant class axioms, derived from an ontology by means of a web services framework. We described the system and its core components; OWL-DL telecom hardware ontology, ontological-natural language processing pipeline, an ontology axiom‐extractor; and the semantic assistant framework.
To respond to customer information requests in a timely manner, contact center
workers need to search through many types of knowledge resources including user
manuals, training, marketing, engineering, and case resolution databases. Companies
face increasing contact center costs as their products and information support services
evolve. Key drivers in this business process are the reduction in number of cases
escalated to more experienced workers and reducing time spent in resolving cases.
Training time must also be greatly reduced and workers with less knowledge should
be able to fill in for others on an ad hoc basis. A knowledge-based approach is
appropriate in this context. Moreover expressive features of the OWL language [
product‐specific OWL‐DL Ontologies that are consumed by a middleware
leverageing web services to push information to client side knowledge workers.
The system architecture for our knowledge sharing paradigm includes the following
core components: (i) Ontology; (ii) Ontological NLP pipeline; (iii) Ontology
Axiom‐extractor; (vi) Semantic Assistant Framework [2]. These components are
integrated into an online annotation workflow shown in Figure 1.
The scope of the ontology included hardware components that comprise the chassis of
a telecommunications switch and more specifically the compatibilities of different
hardware components therein [
The Ontological NLP pipeline is a text mining solution based on open source GATE
framework [4]. GATE included two main steps, firstly a GATE annotation pipeline
for recognizing named entities in source text and connecting them with canonical
names and semantic classes, and secondly JAPE rules [
The Semantic Assistant Framework is a service‐oriented architecture used to enhance existing end‐user clients, such Open Office Writer, with online Telecom related text analysis capabilities provided as a set of web services. The Ontological NLP pipeline links Telecom named entities occurring in a document opened on client side with existing ontologies on server side (See Figure 1). The system annotates each named entity with canonical name, class name and related class axioms providing annotation for documents on the client side. Figure 2 demonstrates an example of client-side annotated text2.1The proposed system is scalable and extensible allowing users to easily customize the information to be delivered as annotations depending on the availability of ontologies with defined axioms linked to canonical names for entities. 1 ANNIE Information Extraction system developed for GATE. http://gate.ac.uk/ie/annie.html 2 We used Open Source Writer integrated with the Semantic Assistant plug-in as a client-side application. Our annotating web services is compatible with any other existing or proposed client-side software that has web services client functionality.
When replacing unserviceable telecom hardware customers phoning a contact centre for technical product support often have questions about which components are compatible with their chassis. Older chassis models of a telecom switch may be compatible with specific models of power supplies and incompatible with later versions. By combining the Semantic Assistant with existing text mining and axiom extraction pipelines we leverage the DL-axioms in our OWL ontology to great effect, rapidly communicating hardware incompatibilities to contact centre agents through their desktop applications. Installing incorrect power supplies can result in equipment failure, injury to the installer, network downtime for tens of thousands of customers.