Case Study

Amiqus - Using Data Science to provide affordable legal services

Posted on September 20, 2016

Amiqus, a legal-tech software company, was set up in Scotland in 2015. Its aim is to provide consumers and SMEs with an affordable alternative to court. With 1 million cases going unresolved in the UK each year due to 1 in 3 people not being able to find or afford legal advice, Amiqus offers free access to legal information through open data.

By creating software to analyse legal information, Amiqus can predict the outcome of claims and disputes to help avoid court. Amiqus wanted to supply this data to individuals and small businesses for free to support and aid their cases.

With support from Scottish Edge, the team at Amiqus went to The Data Lab to help facilitate an academic partnership.

The Data Lab enabled a link between Amiqus and the Computer Science and Law departments of Strathclyde University and provided Amiqus with £35,000 worth of academic time.

After six months of work, Amiqus built a functioning proto-type which they now own. The software analyses historical, open source, court and legal case information using sophisticated data programmes.

Amiqus first considered using natural language processing of this information but adopted a purely statistical approach to process and match the text.

Apache Lucene, a free and open-source information retrieval software library, was considered as an easy option to perform IR-style searching over the text. However, an alternative algorithm was used as this has been established as superior for unparameterised long phrase matching.

Server-side code is written in Java and hosted on Apache Tomcat, developed using the Eclipse development server. Open Government documents such as copies of relevant legislation are published in a stylised form of HTML, which is converted into the W3C DOM Standard using the open-source JTidy, also supported by W3C.

As Amiqus continued to refine and improve their product throughout the process, further possibilities with the software became apparent. For example, by applying sophisticated term-matching at first point of contact, Amiqus could facilitate better search results. In practice this means searches for ‘car’ will return results for ‘motor vehicle’ or ‘flat’ will return results for ‘dwelling house’. Users therefore do not need to be familiar with legal terminology in order to return the best and most relevant results.

In the beginning, Amiqus built a minimum viable version of a larger system to de risk their proposition and really understand the many opportunities their data analysis would present as they grew their scope. Amiqus cleverly decided that they didn't want to create any artificial constraints in tying their work or analysis to a particular infrastructure or software products.

Now, as the Amiqus’ software continues to mature, these anticipated opportunities are beginning to reveal themselves. The team have recently found that Deep Learning may become a viable method of analysis.

As part of their current analysis, Amiqus monitor the mapping between query and results using principles similar to modern search engine technology. These systems naturally work best once huge volumes of mapped data are analysed. As this process becomes more sophisticated, the possibility that these techniques can be adopted through Deep Learning is higher and something Amiqus aims to implement.

Currently in the final stages, this will soon go live on its website (www.amiqus.co). With a current focus on consumer law, Amiqus is planning to expand their legal expertise into employment law. Currently users will be provided with a link to relevant examples of case summaries but Amiqus want to expand this.

The Data Lab continued to bridge the connections between Amiqus and Strathclyde University. The Data Lab’s in-house data scientists supported Amiqus by monitoring the progress and ensuring that Amiqus was able to extract value from data and achieve their business goals.

Mailing List


To recieve updates from The Data Lab please enter your details below.