Text Mining Technologies
Text Mining Solutions use a range of innovative technologies to provide the portfolio of services requested by our clients. We work with a number of technology developers to ensure access to the latest advances in text mining, analysis and visualilsation tools.
TMS has made a major leap forwards in the way end users can search and visualise indexed records. In partnership with GATE, Text Mining Solutions has developed an online search and visualisation tool called TMS Prospector. Documents which have been annotated by TMS are loaded into a Mimir index and Prospector sits over the index acting as an interface through which data can be queried and retrieved.
Users can build and save complex searches using the query editor. Documents which match the search terms are segregated from the archive and are available for viewing or filing. Prospector’s innovation lies in its ability to present search outputs in
visual format – as either a correlation matrix or a correlation chart.
Rapid Evidence Assessment (REA)
The key objective of our rapid evidence assessment (REA) service is to quickly assess and summarise existing evidence about particular issues. Here, for example, we have compiled and processed a set of approximately 4,700 records concerning fetal and neonatal mortality and presented the evidence in this 2D visualisation. PDF download.
The column headings are Diseases and Disorders and the cell colour indicates the number of documents on that subject, red being many and blue being few. By downloading the documents it is evident, at a glance, the main causes of foetal and neonatal mortality are pregnancy complications, suffocation, pre-eclampsia, hypertension, respiratory distress syndrome, sepsis, foetal distress, eclampsia, anemia, anoxia, etc.
Natural Language Processing (NLP)
Text mining is a means of analysing (reading and interpreting) text in order to extract certain bits of information such as pre-specified types of events, entities, or relationships using Natural Language Processing (NLP).
This is typically done either by rules based pattern matching or machine learning and in practice often involves a combination of both approaches to get the best results and maximize performance.
At Text Mining Solutions we have used a combination of rules and machine learning to classify scientific abstracts in the field of health economics and applied the technique to customer surveys in order to understand the expressed level of customer satisfaction by product and service.
Text Mining Over A Range Of Domains
Text mining over a diverse range of domains offers a promising means to draw out possible weak signals.
Intelligence about the future is a key resource for building robust growth plans, anticipating and responding to disruptive technology, and successful innovation. Information is rarely of greater value as that which indicates changes in the dynamics of existing trends or the onslaught of disruptive technology. One long-term objective of many businesses is to take advantage of fragmented, unrefined and low-quality information. Building a “weak signal” interpretative capability is a mater of continuous organisational development in terms of assumption re-appraisal, NPD and innovation capability, collective creativity, and actionable insights. Image file