Search Acumen, the property data insight and technology provider, has introduced two machine learning tools to speed up the legal due diligence involved in property transactions. The UK-based company is using the Amazon Comprehend system as the base platform for the development of its new tools.
Amazon’s Comprehend system is described by the giant tech company as ‘an NLP service that uses machine learning to find insights and relationships in text’.
The first Search Acumen application is an address cleansing tool that it believes will reduce hours or even days of administrative time spent error-checking and correcting property addresses, and will create a more accurate and efficient due diligence process.
Lawyers can then spend more time analysing the data and consulting with clients on any potential risks to progress transactions more efficiently, the company told Artificial Lawyer.
‘Incorrect address data can be a major issue, in particular for complex transactions with hundreds or thousands of addresses, as an isolated error can hold back the entire process until it is resolved,’ they explained.
The machine learning tool reconstructs address data – cross-referencing it against over 30 million official UK addresses – to produce a consistently accurate result. This can then be matched against external records from HM Land Registry, Local Authorities and other key data sources.
As with all machine learning, the tool is continually being fed data, trained and updated, they said. The address cleansing tool is now available to all users of Search Acumen’s Portfolio Cloud service.
Search Acumen has also launched a property search request tool that can categorise emails and attachments and forward data to the correct location.
The technology is capable of filtering and combining the contents of emails and attachments received from Local Authorities and other external sources in one place in seconds. It is being constantly fed information so that in future it can manage the company’s Customer Service mailbox ‘without intervention and at great speed without error’, freeing up the Customer Support team to handle more complex tasks.
Andrew Lloyd, Managing Director at Search Acumen, said: ‘We have long been advocates of AI, machine learning and other transformative technologies to streamline and accelerate vital legal real estate processes. This is the first stage in our transition to becoming a business with AI and machine learning at its core.’
And in terms of how they used Amazon’s Comprehend system, Lloyd explained to this site: ‘Comprehend has a series of default entities like names and places that it has already learned to find in documents. For our project on property searches, we have added to this model with custom entities specific to the search documents we are processing.
‘We have created a training interface that allows us to process documents through Comprehend and identify these additional entities that we require, including the type of search, the provider of the search, the reference numbers, invoice documents, general queries and actual results.
‘This raw data is not particularly end user-friendly so there is a need to present this data in a context that will be more easily consumed by the lawyers’ clients in the same way that lawyers prepare a report on title for the benefit of the consumer.’
For Search Acumen this is a step up in terms of the tech offering they can now field for clients. It’s also interesting that they chose Amazon’s ‘generalist’ NLP system and then trained up additional search capabilities, rather than working with an existing NLP company in the legal and property field. This has allowed them to build their own product, although this has involved plenty of work on their part.
Also, whether giant tech companies such as Amazon will break into the ‘legal AI’ space is a question many in this sector often ask, but for now this is probably the most likely type of outcome, i.e. companies will take the generalist NLP systems already out there and then adapt them to their specific needs, just as many vendors do with other NLP models that have been available in the market for a long time now.
That said, Amazon does have a specially trained NLP set for the medical sector that can read everything from doctors’ notes to medical trial reports. However, they have not yet made any indication they have plans for a specialised tool kit for legal work. If they did, that could change things.