Palimpsest

Flexible data extraction and structuring solution

Palimpsest is a flexible data extraction and structuring solution that can be tailored your organisational needs. It takes your organisation’s unstructured text data and applies a range of techniques to create structured datasets which can then be incorporated into your quantitative analysis. Palimpsest enables the intelligent preservation of the information value embodied within the original text.

Palimpsest is a hybrid approach, utilising and integrating both human and machine analysis. We work with experts both inside and outside your organisation to identify useful structures to impose on the data and infer its deeper meaning, and exploit machine learning to apply these human-derived insights at scale. Palimpsest delivers bespoke datasets which are structured to capture the wealth of information contained within natural language text, and which map directly to your decision-making.

The Palimpsest approach uses automated extraction and other NLP techniques in combination with expert annotation and the use of innovative crowdsourcing methods to produce datasets. The use of machine learning algorithms to assign confidence to the annotations, means that you can be sure the data has been accurately and validly structured.

If you are interested in understanding how Palimpsest can help you unlock the potential value of unstructured data, please get in touch.
[email protected]

1Understanding the data challenge

Knowing what decisions you need the dataset to support is crucial to structuring the data appropriately.

2Schema design

We work with you to create the most useful and relevant data structuring frameworks.

3Automated extraction

We apply a range of tools to begin applying structure to the data.

4Expert annotation

Our approach employs human analysts who understand the problem and the organisational context to help interpret the structuring framework.

5Crowdsourced structuring

We utilise crowdworkers to get human input on the deeper meaning of natural language text.

6Dataset analysis

This involves the examination and validation of the datset\'s content, providing summary data about its quality.

7Model integration

This involves ensuring the dataset is in the right format to be incorporated into your organisational analysis.

8Informing the decision

We present the insights arising from the newly structured datasets in a way which maximises their decision relevance.