What Is NLP and Why Should Lawyers Care?

Will computers ever replace attorneys?” We’ve heard this question, in one form or another, for over a decade. And the answer is still “No.” Or at least, “Not for a while.”

Technology innovations are changing the competitive landscape, however, and they’re creating new opportunities for lawyers and legal professionals. One area in particular is called Natural Language Processing. And it’s coming your way.

Natural Language Processing

NLP is the application of linguistics, statistics, and computer science, to problems related to spoken or written language.

When you begin typing into Google, and Google predicts the rest of the phrase, that is NLP at work. When you ask Siri for directions, and “she” “understands” your request, that is NLP at work.

Examples of NLP applications to solve other computational problems include:

  • Information retrieval: Given a query, find “relevant” information—i.e. information that the user probably wants to see, even if they weren’t able to clearly articulate their need.
  • Information extraction: Given an audio recording, or a lengthy text document, extract facts (for example: people, places, events, or transactions) and store them in a database.
  • Speech recognition: Recognize a spoken sentence or phrase, and convert it to text.
  • Question answering: Interpret a question, and search a set of documents to find the right answer.

NLP is a mature field, and it is being used in new and creative ways across a wide range of industries. Many other applications beyond this list of examples use NLP.

From Information Retrieval…

In the legal field, NLP has been used in various ways in ediscovery for years. From an NLP standpoint, ediscovery is primarily an “information retrieval” task: helping teams search for documents relevant to the legal discovery process. NLP is used to separate and identify “relevant” documents (i.e. documents relevant to a given query) from a larger body of documents.

Simpler ediscovery applications allow users to search for documents by keyword, phrase, or using a “bag of words” approach that simply looks for documents that share a lot of words in common with the search terms.

More advanced applications support “concept searching.” For example, if you’re searching for documents related to the concept of “payment,” a concept search will find documents that contain related words, such as “pay,” “invoice,” “compensation,” “fee,” and so on. This type of search typically finds relevant documents that would be missed if you only used a keyword search.

Some ediscovery applications support predictive coding. Predictive coding allows users to “train” the application by providing samples of documents they consider relevant. The application then searches for similar documents and presents them to the user, who either confirms them as relevant, or dismisses them as irrelevant. By repeating this “train-search-review” process until the user is satisfied with the results, thousands (or millions) of documents can be reviewed in a fraction of the time as compared to traditional techniques.

… To Information Extraction…

Legal technology applications are now shifting towards “information extraction“: applications that search through unstructured text, and extract useful data. These applications span many areas of contract law and case law:

  • Contract management departments are using NLP to find and extract key terms from their contracts. They’re able to create reports that summarize terms across contracts, and compare contractual terms to “standard” terms for risk assessment purposes. These applications automatically extract dates, dollar amounts, and other key information for planning, budgeting, and risk mitigation purposes.
  • Derivatives traders are using NLP-powered software to analyze derivatives contracts. This software extracts interest rates, termination events, and other relevant information. Once data has been extracted from the agreements, it is used to support trading decisions, manage collateral, and support regulatory and compliance requirements.
  • Law firms in the energy industry are exploring the use of NLP to speed up oil and gas title abstracting. For energy clients with hundreds of parcels of land, oil and gas title abstraction can take months to identify and summarize all conveyances and encumbrances. Using NLP to identify and extract key information, it may be possible to reduce the abstraction process from months to weeks.
  • IP attorneys are using Lex Machina to extract key data from public court records—parties, patents, outcomes, and other relevant data. Lex Machina links this data together to create summary reports that help users craft IP strategy and win cases.

All of these applications use information extraction techniques to pull useful information out of text. This information can then be used in reports, predictions, and other tools to help lawyers and legal professionals to make better-informed decisions – in a fraction of the time – and to discover insights buried within hundreds (or thousands) of documents.

… To Semantic Understanding.

Many NLP applications today use “shallow NLP” or “statistical NLP” techniques to achieve their goals. These applications don’t “understand” the meaning of a word, phrase or sentence—they’ve just memorized certain words, patterns, or statistical associations between words.

For an application to show “intelligent” behavior (for example, to interpret a legal contract), the application would need to understand legal concepts, and it would need to have some way to reason or combine information from different sources. NLP researchers are working in two areas—semantics and pragmatics—that may eventually enable computers to “understand” text in a meaningful way. While still a ways off, this research may someday lead to the development of applications that can acquire knowledge on their own, or reason in a robust, “intelligent” manner. If this occurs, attorneys could potentially use these applications to reason about contracts and cases, make predictions about potential outcomes, or research topics under human guidance.

Although we won’t be seeing robot lawyers any time soon, NLP is a useful tool that is opening up new opportunities, and changing the competitive landscape. Today, law firms and legal professionals are using information extraction applications to make better-informed decisions—in a fraction of the time—and to discover insights buried within hundreds (or thousands) of documents. Tomorrow, attorneys may use NLP-powered applications to reason about contracts and cases, make predictions about potential outcomes, or research topics under human guidance.

Regardless of your practice, NLP will change the way you work—and probably within the next few years. Stay informed, and stay ahead of the curve—this is an exciting time!

About the Author

Lars Mahler is chief science officer at LegalSifter (www.legalsifter.com). He can be reached at lars@legalsifter.com.

Send this to a friend