5 tips for developing an ontology

Posted by Annabel Snow, Consultant on 28 May 2015

Tags: ,

While information professionals have been designing file plans and thesauri for decades, chances are ontology development will be a new endeavour, created in a digital paradigm to deal primarily with digital content. Before grabbing a tool and collecting every word ever written, here are some key points to think about when developing an ontology: 

1 - Be clear about the use for the ontology.

When you start looking at the ontology, have a clear idea of what you want to do with it; what outcomes do you want to achieve from developing it?  There is a difference between developing a conceptual ontology that is an authoritative source of terms (independent of any actual use or system), and developing one for a specific purpose such as auto-classification of content or to assist business intelligence.

You could have an ontology that serves both purposes but you need to be clear on outcomes so that you know which terms are being defined for ‘interest’ and which ones are then available to other tools for a business purpose. Understanding this can then drive both your ontology development and classification strategy. 

2 - Start small but think big.

The practicality of developing an ontology is such that you generally choose an area and build it out, one section at a time.  However as mentioned in point 1, you need to have the end goal in mind; so scope appropriately and ensure that your modelling covers the entirety of the organisation, even if you deliver in phases. 

An added bonus of taking an enterprise-wide approach in the scoping phase is that it’s also a chance to set the wheels in motion for the establishment of governance structures and maintenance processes (see point 3).

 3 - Ontology is part of your information management framework

Your ontology doesn’t exist in a vacuum - And as such, it needs to come under the appropriate governance framework.Start building governance structures and maintenance processes from the outset. The organisation needs to consider, for example, ‘who is going to take ownership of the ontology? and ‘where does responsibility for its on-going use and maintenance sit?’

If the ontology and its management sits outside of your overarching information framework you run the risk of it not being maintained, its on-going value to the organisation could be lost over time and there could be a loss of integrity and trust in any related systems.

Having these frameworks in place will assist in keeping your ontology responsive to changing internal language, terminologies, and customer expectations and experiences.

4 - The process needs to be iterative

This seems like an obvious point but it is critical that the ontology (whether it is being used for classification or not) is consistently tested against real life examples, such as organisational content or schema. One of the core benefits of having an ontology is to aid the findability and usability of content by your organisation, customers or users. To ensure this experience is consistent and reliable you need to invest in testing and ensuring your ontology is up to date.

Testing also includes consultation and interaction with stakeholders both internally and externally, where possible. This assists with its development but also ensures the model does not become completely disconnected from its source or end-user.

Most importantly, the importance of ongoing evaluation and feedback generation should be built into any timeframes or development plans as it can be easily overlooked, and the time it takes can be sorely underestimated.

 5 - Don’t reinvent the wheel

Where possible draw on existing information classification standards, like ISO 3166 for country codes, or check if there are industry-wide terminologies or approved taxonomies that you can leverage. Standards will not necessarily address on-going maintenance and control (refer back to point 3). And don’t be surprised if there is some data cleansing required to fit the purpose of the ontology, it’s going to give you a base to work from.

Ontology development for your organisation

We hope that you have found these tips useful, here’s one final word of advice for you - no client ever starts with a blank canvas. If you look around you, you’ll soon see that there are a number of different models already employed; your intranet site structure, records classification scheme, vocabulary etc. It’s knowing how to harness this to address the overarching information management goals of your organisation and what the right tools are to make this as easy as possible.

ontology blog 3

 

Making information management easy

Our team can help you make sense of content classification. We can support you in the development of a valuable and useful ontology but also help you realise the benefits of this approach to information management.  If this blog has piqued your interest, get in touch – we’d love to talk in more detail with you about developing the right solution for your business’ data management.

Further reading

-          Overcoming the pitfalls of ontology authoring

-          Ontology to support web design activities in e-commerce software development process