Regulytics

Regulytics® is our web app for the automated comparison of regulatory texts on paragraph level. The main goal is to make gap and impact analyses more efficient by identifying semantic similarities between different regulations.
The app does not provide cumbersome general reports, but concise, tailored and immediate information about content-related similarities.

Financial service providers can now request a free trial account – preferably via email.

Our solution can be accessed by authorized users via this link.

The motivation and background for this solution are explained here.

On request, we grant financial service providers a free trial account for a limited period of time and information for further use. In addition, we offer an enhanceable offline version with additional functionalities for analyzing internal texts.

In addition, we offer a web app to analyze the building regulations (‘Bauordnungen’) of the various German states as a free demo version of Regulytics: link

Usage

The solution is easy and straight-forward to use. Languages as well as start and target regulations can be selected via dropdown menus. Another input is the number of topics for the natural language processing model (we recommend values between 100 and 500). Based on these inputs, the overall similarities are calculated and displayed.

Overall similarities

After choosing a “start paragraph” in the start regulation, the complete similarity matrix for all paragraphs is calculated.

Similarity matrix

Also, the top 10 most similar paragraphs (of the target regulation) to the start paragraph are displayed. Also shown are the  paragraphs of the start regulation with the highest and lowest similarities to the target regulation. Thus, overlaps as well as changes between regulations can be easily detected.

Similar and different paragraphs

Important note When determining similar and different paragraphs, the quantitative cosine similarity measure should always be kept in mind.
E.g., if the most similar paragraph to a particular paragraph has a similarity of less than 0.5, it can be assumed in most cases that there is no actual similarity, i.e. no similar paragraph exists.
The same applies to the determination of differences.
A determination of lower (for similarities) and upper cosine similarity bounds (for differences) is possible and sensible, but depends on the respective question. We are happy to assist you with this question.

All results can be downloaded as XLS files.

Contact

Dr. Dimitrios Geromichalos
Founder / CEO
RiskDataScience UG (haftungsbeschränkt)
Theresienhöhe 28, 80339 München
E-Mail: riskdatascience@web.de
Telefon: +4989244407277, Fax: +4989244407001
Twitter: @riskdatascience

Regulytics® is a registered trademark at the German Patent and Trademark Office.