2. Ontology Learning
Knowledge is considered to be “the information needed to make business decisions”1, and so knowledge management is the “essential ingredient of success” for 95 per cent of CEOs [Manchester 1999]. A company’s value depends increasingly on “intangible assets”2 which exist in the minds of employees, in databases, in files and in a myriad documents. Knowledge management technologies capture this intangible element in an organisation; and make it universally available. The most widely used method of mapping the knowledge of a domain is to use an ontology describing such a domain. Ontologies can act as an index to the memory of an organisation and facilitate semantic searches and the retrieval of knowledge from the corporate memory as it is embodied in documents and other archives. There are many real-world examples where the utility of ontologies as maps or models of specific domains has been repeatedly proven.
Human
Introspection KNOWLEDGE ELICITATION
Human Introspection KNOWLEDGE BASE DEFINITION
Human
Introspection KB POPULATION Figure 1 A basic knowledge base development cycle (current reality)
For knowledge to be managed it must first of all be captured or acquired in some useful form, e.g. stored in an ontology. Knowledge acquisition (KA) is a complex process which traditionally is extremely expensive. Yahoo currently employs over 100 people to keep its category hierarchy up to date [Dom 1999]. An appropriate model of knowledge acquisition under the current paradigm would note that it is 1 Philip Crawford, European Vice-President of Oracle Corp. cited by [Philip 1999] 2 A term coined by the industry consultant Karl-Erik Sveiby an entirely manual process. The typical process of KA is outlined in Fig. 1: initially knowledge is elicited either from a set of documents or from an expert of the domain. Then an ontology is defined, finally a set of instances are associated to concepts in the knowledge base. It is widely agreed that the key factors which impede the wider use of ontologies both in research and commercial applications are time, cost and subjectivity. Time and money committed to ontology development is substantial as the years of development at Cyc have shown and the manpower requirements of Yahoo prove. Subjectivity is inevitable, given any method which depends on individual introspection or elicitation for data collection.
Although efforts exist to automate KA most of these are isolated to specific aspects of the construction cycle, little effort has been spent on creating a comprehensive set of tools for KA. In this paper, we propose a set of techniques to largely automate the process of KA, by using technologies based on Information Extraction (IE), Information Retrieval and Natural Language Processing. We aim to reduce all the impeding factors mention above and thereby contribute to the wider utility of the knowledge management tools. In particular we intend to reduce the introspection of knowledge engineers or the extended elicitations of knowledge from experts by extensive textual analysis using a variety of methods and tools, as texts are largely available and in them – we believe – lies most of an organization’s memory. Ontology construction aims to capture knowledge in a usable format. The nature and granularity of the ontology depends on its eventual use; typically IS-A hierarchies are central but other emphases are often encountered.
The process of ontology construction may be divided into three stages the first two of which contribute to the learning of the ontology structure and the third is used to populate the knowledge base with instances. These stages are illustrated in the rest of this section. refinement [Järvelin and Kekäläinen 2000]; iii) passed on to the next development stage below.
We propose to introduce automation in the stage of
taxonomy construction mainly in order to eliminate or
reduce the need for extensive elicitation of data. In the
literature approaches to construction of taxonomies of
concepts have been proposed [Brown et al. 1992, McMahon
and Smith 1996, Sanderson and Croft 1999]. Such
approaches either use a large collection of documents as
their sole data source, or they can attempt to use existing
concepts to extend the taxonomy [Agirre et al.2000, Scott
1998]. We intend to develop a semi-automatic method that,
starting from a seed ontology sketched by the user, produces
the final ontology via a cycle of refinements by eliciting
knowledge from a collection of texts. In this approach the
role of the user should only be that of proposing an initial
ontology and validate/change the different versions
proposed by the system. We believe an ontology
construction method should a) permit multiple placement of
terms in the structure, b) allow rapid recalculation of the
structure, c) provide monothetic labels for nodes, d) allow
the input of seed ontologies for further expansion.
We intend to integrate a methodology for automatic
hierarchy definition
The advantage of this integration is that as knowledge is continually changing, we can reconstruct an appropriate domain specific ontology very rapidly. This does not preclude incorporating an existing ontology and using the tools to extend and update it on the basis of appropriate texts. Finally an ontology defined in this way has the particular advantage that it overcomes the well-known ‘Tennis problem’ associated with many predefined ontologies such as WordNet, i.e where terms closely related in a given domain are structurally very distant such as ball and court.
In addition we intend to employ classic Information Extraction techniques such as Sheffield’s named entity recognition system [Humphreys 1998] in order to preprocess the text, as the identification of complex terms such as proper names, dates, numbers, etc, allows to reduce data sparseness in learning [Ciravegna 2000].
We plan to introduce many cycles of ontology learning and validation. At each stage the defined ontology can be: i) validated/corrected by a user/expert; ii) used to retrieve a larger set of appropriate documents to be used for further
This stage proceeds to build on the skeletal ontology in order to specify, as much as possible without human intervention, relations among concepts in the ontology, other than ISAs. In order to flesh out the concept relations, we need to identify relations such as synonymy, meronymy, antonymy and other relations. We plan to integrate a variety of methods from the literature, e.g. by using recurrences in verb subcategorisation as a symptom of general relations [Basili et al. 1998], by using Morin’s user-guided approach to identify the correct lexico/syntactic environment [Morin 1999], and by using methods such as [Hays 1997] to locate specific cases of synonymy.
Once the ontology has been learnt, there is the problem of
retrieving instances in order to populate the resulting
knowledge base. This is a key issue, in order to use the
ontology as index to the organisation’s memory, for
example by allowing semantic searches and the retrieval of
knowledge from the corporate memory. In many
environments KB population is performed manually by a
user via instance identification in a text corpus. We plan to
automate this process as much as possible by using a
combination of text classification (TC)
Knowledge is only of value when it can be used effectively and efficiently. The management of knowledge is a key element in extracting its value. In this position paper we have outlined how we are addressing the issue of automating the Knowledge Acquisition process in order to reduce both required time and cost of KA, and subjectivity in the resulting ontology. Overall we believe, this will make knowledge management not only more acceptable in a commercial environment but also contribute to the overall productivity of the economy.
The work outlined above is being undertaken by the University of Sheffield in the context of AKT (Advanced Knowledge Technologies, http://www.aktors.org), a multi-million pound, six-year project involving the University of Southampton, the Open University, the University of Edinburgh, the University of Aberdeen, and the University of Sheffield. AKT will extend knowledge management technologies to exploit the potential of the semantic web, covering the use of knowledge over its entire lifecycle, from acquisition to maintenance and deletion. It began in October 2000 and will comprehensively addresses six main challenges, which are fundamental bottlenecks to knowledge management: • acquisition • reuse • modelling • publication • retrieval/extraction • maintenance The work at Sheffield will on provide a library of Natural Language Processing based tools for different types of knowledge management tasks.
This work is supported under the Advanced Knowledge Technologies (AKT) Interdisciplinary Research Collaboration (IRC), which is sponsored by the UK Engineering and Physical Sciences Research Council under grant number GR/N15764/01. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing official policies or endorsements, either express or implied, of the EPSRC or any other member of the AKT IRC.