Seeking words with meaning
15 Novemer 2011
The analysis of textual information from surveys, online sources and client databases continues to be a hot topic for research professionals.
In the current economic climate, to get more from less it is important to be able to analyse responses from a wide range of qualitative feedback mechanisms. For some time, different forms of text mining and analysis tools have been used in mainstream customer, brand, employee and social research. These have included solutions such as first and second generation natural language processing (NLP) software as well as various coding tools.
The text mining revolution
Over the last six years, ORC International has been integrating a number of text mining methodologies into its research studies. The solutions that are now offered to clients are representative of a company that is driven to deliver innovation through cutting-edge technology.
Text mining can help our clients in a wide range of areas, for example:
- Analysing verbatim responses collected via quantitative survey questionnaires
- Examining transcript, forums and reports from qualitative research
- Analysing feedback received from consultation exercises.
It also presents the opportunity to analyse open feedback from less structured sources such as CRM or contact centre notes, complaints and customer suggestions. The software can be used in both qualitative and quantitative studies, but is most powerful when integrated into large-scale studies with an extensive amount of free-form, unstructured feedback.
On these larger projects, the ability to automate tasks like cleaning, categorisation and analysis can be very useful. As well as making the process smoother, it can drive costs down and improve turnaround times after the initial set-up and calibration phase.
A four stage text mining approach
Importantly, our approach to categorisation enables us to analyse the open comments in a much more detailed manner and lends itself to further statistical analysis using correspondence and correlation techniques.
Giving the people a voice
In the public sector field, ORC International carried out text analysis on the Your Freedom website; a site set up by HM Government to allow the public to respond to laws and regulations. It asked the public to indicate specific laws they thought should be repealed and why. The very nature of the site was comments driven, allowing the team at ORC International to truly test the scope and functionality of the text mining software.
Over 100,000 comments were submitted through the site. The data mining tool enabled analysts and statisticians to dissect the individual responses, analyse keywords, phrases, themes and sub-themes. Based on relevance and insight, the software gathered a subset of 25,240 comments to be reviewed further.
The comments were then all categorised into three themes, and then further sub-themes, giving the government a clear view of the key issues and topics being discussed.
The top three categories were crime, health and rights. As an example, within the crime theme the top areas identified were cannabis legalisation, imprisonment, control orders and drugs generally. And within the health theme, smoking was by far the most discussed topic.
Tomorrow’s world
The analysis and reporting of textual research data is a complex process and benefits from constant development and advances. We are committed to our offering as technology continues to evolve. When developing our current capabilities, we have tackled a range of issues with management and processing, and over the next year we are striving to make continued improvements to areas including sentiment extraction and analysis, the accuracy of NLP comprehension and multiple language processing.
