We are living in the era of digitalisation. Today, there is an increasing demand for digitising our physical environment, to transform traditional businesses into their digital counterparts, to bring new added values to business by the use of many sources of information in a digital form.
We want to convert our world into a digital representation which should allow us for its easy manipulation, interaction, as well as to be able to get new type of insights and in the end be an extension of our intelligent capabilities.
By seeking that goal, we are constantly increasing, in an exponential fashion, the stored and available data around us. Data comes in many distinct flavours and sources: text, sound, signals, images, videos; IoT devices, computers, mobiles, stock market, medical devices, etc.
So, the era of digitalisation brings us to the era of Big Data and the use of cutting-edge technologies related to machine learning and artificial intelligence. Those tools are needed in order to analyse and make use of such an amount of data.
Many of the information being digitised comes in the form of text. There is already a data deluge when looking at the text data being stored: documents, twitter, blogs, social networks, websites, chats, news articles, etc.
In parallel, there is an increasing need for mining those text data sources in order to squeeze and extract their hidden insights that would be of great value when taking business decisions, anticipating issues and being combined with other data types to extract other insights.
A typical traditional area affected by the digitalisation is to be found within the legal sections of many big companies.
There is a huge trend for digitising historical records, documents, contracts, correspondence, etc. Besides, there is also a need to simplify and automatize business processes and bring new searching and smart capabilities to the legal departments. As a consequence, the number of text documents being stored every day increases dramatically.
Data deluge, increasingly tougher regulations, compliance, contract constraints and clauses, complexity of contracts (links, versioning), litigation etc. make review processes by legal experts very costly and time consuming and also error-prone. All of that, together with additional risks can make business to get into serious situations.
Rapid changes in regulations make the whole story even worse and new challenges are hence presented.
We need more indexing text functionalities but also smart automated processes taking care of text data mining, information discovery during litigations, text summarization, topic classification, etc.
In those challenges, computers are best suited when trying to find complex patterns, discover new insights, relationships and provide more relevant information to the business to allow them to take informed decisions.
We, at Swisscom, are faced with the same data challenges. Legal users consume many hours of their work time every year when involved in reviewing, litigation processes, etc.
Regulations are the more and more complex. Hence, many hours, even weeks are currently needed in order to find the right information amongst the huge number of text pages.
To make things even harder, legal documents reside on different sub-systems, saved as scanned pictures, stored with different metadata, being redundant, incomplete, noisy and even (near-) duplicates and versioning of same contract or legal document can be found.
New cutting-edge technologies in the realm of Artificial Intelligence and Machine Learning are being applied to solve these new modern challenges with astonishing success.
Those machine learning techniques are able to spot very hidden patterns when presented with real data. They are best than humans in doing so and they are less prone to errors.
Insights revealed from those machine text analyses will allow legal users to take quick and informed business decisions for the benefit of their companies.
At Swisscom, we decided to bring those new technologies and smart functionalities to the daily life of a legal business user.
So, the first thing to do is, to understand how legal stakeholders work, what business processes take place, what problems are currently occurring and which ones are foreseen in the near future.
Besides, we also need to understand the sort of issues they are encountering in their day-to-day business and how technology could help them in a smart way.
We defined several goals for a first (AI-)PoC for legal matters.
The benefits associated to this project not only rely on the above functionalities but also on the competitive advantage of having cutting edge technologies at the service of the legal departments.
Besides, because of this project, optimized business processes are now in place and all the available documents are in a numeric form and be accessible in a secured and authenticate manner from a single system.
On top of that, (near-) duplicates, versions of same document have been found, and also unexpected patterns and other insights completely unknown to the legal users are now available.
We will be soon starting the new phase of the project, in which we will be bringing more smart functionalities thanks to the use of artificial intelligence, machine learning and natural language processing technologies.
The use of these technologies, mainly the ones related to text machine representations and deep learning, can also be applied to many other business processes in regards of text analytics and text mining.
Finally, just to whet your appetite, I could mention some of the envisaged use cases: