Legal Contracts Services Platform
The client is a leading software company providing enterprise SaaS that leverages machine learning to accurately extract critical information from legal contracts.
The company mainly focuses on helping organizations to maximize profits and minimize risks in their commercial relationships by understanding their legal business contracts.
The need for maintaining and enforcing business contracts has grown in the past decade, owing to frequent regulatory reforms.
Because contract documents are increasingly verbose and detailed, they are overwhelming and difficult to comprehend.
A leading software firm wanted to simplify this process by helping companies understand and manage their business contracts through automated annotation and summarization of contracts. Owing to inadequate results obtained by a previous vendor, the project was handed over to High Peak.
CHALLENGES for Legal Contracts
The project was partially developed by another vendor and comprised a lot of bugs. A major challenge for High Peak was to evaluate the already existing design and make significant improvements to the application. The team also developed new applications with more features and better technologies.
Another challenge was to improve the speed of annotation of contracts. The team at High Peak had to gather data to be fed to a machine learning model for training.
The team also had to commit to a UI design that was simple and user-friendly to enable attorneys to use and understand the application with ease.
SOLUTIONS for Legal Contracts
High Peak Software operates as the client’s offshore development center and has assisted in the development of the legal contracts management and annotation system. It comprises:
Automated document identification
The platform uses machine learning models to automatically detect the various types of contract documents when they are uploaded to the platform. For instance, if the document uploaded to the platform is a non-disclosure agreement contract, or a licensing contract, the platform is capable of identifying it as such automatically.
High Peak uses machine learning models that are intelligent enough to learn from the data they consume and more accurately identify document types with time.
Automated data extraction
High Peak implemented automated data extraction for image files by employing Optical Character Recognition (OCR) technology.
The platform allows the user to upload any kind of document including text, pdf, images etc. The textual content in some of these document types are not readable by the machine because they are images. In order to extract information from such document types, the platform uses OCR technology and retrieves relevant information.
This extracted data is then accordingly categorised by our ML algorithms. For example, the algorithms can understand the extracted information presented in a DD/MM/YYYY format as a date.
Automated configurable annotation tool
The High Peak team designed and developed a configurable annotation tool using machine learning algorithms that are capable of identifying the document type and predicting annotations by extracting key-value pairs.
The ML model learns to identify and label annotations within the document. For instance, start and end date of the contract, party names, monetary value, and so on.
In addition, the user can search for a specific value within the document using keywords and create an annotation manually.
Intelligent document summarization
High peak designed and developed an intelligent document summarization tool integrated within the platform. This tool is capable of identifying important sections of text in a contract document and summarizing the information in a way that can be easily understood and consumed by a human intelligently.
Document summarization contains different important highlights of the document categorised by annotation types such as dates, events, and so on. Once the user clicks on a particular category, the annotated segment belonging to that category is pulled up and presented for the user’s view.
The documents identified, processed, extracted, and annotated by our intelligent machine language algorithms are manually verified and validated by attorneys in order to ensure they are highly error-free and fool-proof.
Intelligent search management system
High Peak integrated the platform with an intelligent search management system that includes a global search and local search feature.
The platform uses elasticsearch to enable the user to search for documents by type, by date and so on. In addition, the user can also search for certain elements within the documents as well.
User management system
The Legal Contracts Services platform serves three kinds of users: tagger, reviewer, and publisher.
The tagger annotates the uploaded contract document. The reviewer is responsible for reviewing the annotated contract and to ensure that the annotation is accurate. The publisher is in charge of publishing the annotated contract document.
Because the above-mentioned three different types of users have different requirements, the system must be able to adapt to the same. Therefore, the High Peak team built the platform in a way that caters to these different requirements, with role-based, secure information access.
Content management system
High Peak developed a content management system that is integrated within the platform for secure storage of contracts and other relevant documents. These documents are categorized according to the contract type.
This content management system also enables the manager to assign attorneys based on the type of contract document received from the customer portal (featured in the next section). The manager also regulates the key considerations that attorneys must include in the annotated contract document.
Customers can upload their contract documents on the platform using Google Drive or the file upload option from their device.
Depending on the availability of attorneys and the uploaded file size, the users get a processed contract document which is annotated and summarized.