Simplification of document analysis and comparison with AI driven semantics

Audio Blog Version:

This blog will explain how AI driven semantics is becoming a must to consider when you are a business that has a large volume of documents that require a level of manual review. Imagine the impact to your business if the time you took was reduced by 40% or imagine the cost reduction to your business of not getting things wrong. AI driven semantics is driving this agenda.

AI-Driven Analytics simplifies analysis and comparison of documents in a way that can be difficult for humans to accomplish on their own. It uses machine learning to perform complex analysis on plain text files or PDFs, extracting key elements from them, then comparing those elements with each other. So, if your challenge is too much text and not enough time you need to read on.

What is semantics and why is it important?

Semantics is described as the root meaning of text within any given document or speech. A complex document such as a contract will be open to misinterpretation where the meaning could be obfuscated within layers of overlapping clauses. Misinterpretation could be costly. This is where semantics comes in. The meaning and how it relates to the client are at stake. Usually, only the client can interpret the true essence within a contractual term – but only if they have the facts (and meanings) clearly laid out in front of them, and this can take a considerable amount of time, cost and effort.

How AI driven semantics is driving simplification

The considerable volume of documents created in some industries demands sophisticated tools. Document research and management systems are now using strategies based on machine learning to classify, filter and extract context-based content. These systems help users identify relevant structured portions of text semi-automatically – such as contract clauses. However, while knowledge management systems can deliver automated detection of matching text sections, they often fail to identify meaning – or the change of purpose.

The application of artificial intelligence (AI) to the problem of identifying and comparing semantic meaning is here. Companies such as thingsTHINKING from Karlsruhe have developed technologies that can already identify and compare semantic meanings within complex documentation. Their semantha® platform delivers fully automated semantic processing for legal, contractual, requirements and complex documents. It can be applied to many use cases such as CV matching, NDA analysis, contract review, contract renewals, international contract translation, tender analysis, knowledge management and much more. Users can upload multiple documents to an interface for analysis. These documents can be in different formats (Word or PDF) and various languages. The user can then run a comparison or search against the records to find and identify matching sections of text that imply similar (or the same) meaning. It makes these comparisons and identifies matches in semantic meaning despite different wording and terms in each document. Just as importantly, the semantha® platform quickly identifies if a close match is not available between documents – helping users to discover if an expected intention in meaning is missing from within a document.

Identifying meaning vs application of viewpoint

For many organisations it is not enough that your platform can quickly identify the presence of an expected item of text with specific meaning. We might already know that a phrase or clause is essential – and it is significant that the platform identifies the presence of this phrase – but what about a client’s specific viewpoint of that phrase? For example, a clause specified within a document may stipulate a penalty. Our client might believe that this penalty is unacceptable – and we need the AI to help us identify and highlight the inclusion of this unacceptable condition. In this case, the user needs to be able to identify a particular phrase is ‘Bad’. In addition, the user can apply a ‘Good’ phrase – so anytime your solution finds an item of text similar to the ‘Bad’ phrase, it will suggest or use the ‘Good’ alternative.

By applying your alternative phrases to a document, your solution will learn the user’s viewpoint and apply these learnings to any records available to the platform.

Key drivers behind choosing the right solution

So far, we have talked about comparing, searching, analysing of documents and how AI driven semantics can save you time and, in some cases, up to a 40% reduction of the time you take with manual review. Also, we have discussed some other must haves within the solution.

What are other things you need to think about? We have already talked about how documents can be in different formats. Other things we think you need to consider:

1. Is the solution out of the box? In most cases when you are looking at anything that is AI driven you find it needs to have a period of training. However, there are now solutions like semantha® that are out of the box due to the AI engine it has behind the solution. This means there is no long wait, that can be months with some solutions, to train and implement the model.
2. Capex v Opex and ROI Impact – As an organisation adding a solution to its process automation infrastructure one of the key financials you will look at is Capex versus Opex and the impact it has on ROI. With semantha®, as there is no training, the cost of it can all be seen as Opex and also you start your ROI from Day 1.
3. Languages – We talked very quickly about languages above. When you are dealing with documents you may receive these in a number of languages and it is important that you also deal with these in the same way as other documents. Here it is key that your solution can compare and analyse documents even if there are in different languages.

This is an exciting time for AI driven semantics. As we move toward a more automated future with smarter solutions, we can expect to see more and more use cases for semantic comparison and analysis. As we have seen there are already solutions we have talked about that are driving this agenda with their “plug and play” type approach.

As organisations become more confident in the accuracy of AI driven semantics platform, they can rely increasingly on automation and AI to deliver accurate results far more quickly.

Audio Blog Version: