Press

@ET-DC@eyJkeW5hbWljIjp0cnVlLCJjb250ZW50IjoicG9zdF90aXRsZSIsInNldHRpbmdzIjp7ImJlZm9yZSI6IiIsImFmdGVyIjoiIn19@

by DIEGO MALDONADO ROSAS

@ET-DC@eyJkeW5hbWljIjp0cnVlLCJjb250ZW50IjoicG9zdF9kYXRlIiwic2V0dGluZ3MiOnsiYmVmb3JlIjoiIiwiYWZ0ZXIiOiIiLCJkYXRlX2Zvcm1hdCI6Ik0gaiwgWSIsImN1c3RvbV9kYXRlX2Zvcm1hdCI6IiJ9fQ==@

What are they?

The legal expert systems are computer programs that, feeding from a database of technical information prepared by experts in a subject and ordered by a system of rules of inference, provide information with specific solutions to legal inquiries. In other words, an expert system is software that answers questions of legal relevance.

As examples of successful expert systems we can mention Turbotax, which offers advice, advice and tax declarations for people through its web application, Project CALC, which provides an automatic assistant on the constructibility regulation for architects while designing in Autocad.

In other words, an expert system is software that answers questions of legal relevance.

What elements make up a legal expert system?

Legal expert systems are composed of three elements: a graphical interface, a knowledge database, an inference system.

The graphical interface, also known as GUI – Graphical User Interface – corresponds to what is seen specifically on the screen of the program and this is where the user interacts with the expert system, both by inserting information and receiving answers.

The entry of information can adopt the structure of a digital form, a succession of questions and answers, or a chatbot (basically a chat in which a robot answers you). The same graphic interface will then guide a user on the applicability of certain standards of certain to a specific case.

The biggest challenge of graphical interfaces is to be humanly intuitive and didactic. They should be designed taking into account the knowledge and psychology of the user and should perform the indicated questions allowing these to be answered ideally by means of selectable predefined arguments, numerical answers, or dynamic fields.

The legal knowledge database is the source of substantive information that feeds the legal expert system. Depending on the modality adopted by the inference system, knowledge may be stored – for those systems based on rules – in phrases, paragraphs and predefined sentences. For those inference systems based on statistical methods, data analysis and artificial intelligence, the legal knowledge database will take the form of contracts, court decisions, or other bodies of text that will depend on the matter dealt with by the expert system and will be the power source of the inference system.

The knowledge database is, in other words, the substantive content provided by the legal expert system. Thus, for example, a system programmed to determine the jurisdiction and court competent in matters of litigation will be fed by a series of procedural rules and pre-written phrases, some of which will be communicated to the user when formulating the questions and, those relevant to the problem will reveal the answer to the user’s legal question.

The quality of a legal expert system depends, to a large extent, on the quality and clarity of the substantive knowledge that has been programmed in it. In this sense, the challenge for the legal professional is to know how to correctly translate legal knowledge into logical sentences capable of being understood by software.

The inference system is the heart of the expert system as the logical engine that will generate the answers from the knowledge database and provide relevant content to the user. Inference systems can generally be divided into two groups: On the one hand, systems based on predetermined rules and IF statements and on the other hand systems structured on the basis of data analysis using statistical tools or artificial intelligence.

Inference systems based on rules and IF statements.

An IF STATEMENT “is basically a statement of a program that executes an action depending on a condition. If “A”, then “B”.

The systems programmed based on rules reach certain conclusions from a set of information inputs and their corresponding alternatives provided in the code.

Due to its deductive logical structure, expert systems based on rules can be represented as a tree whose trunk of central questions is divided, according to its variables, into branches of possible solutions and responses. This makes them relatively simple to program and they mainly serve to work with binary or quantitative information sources whose existence and magnitude are not debatable nor depend on the normative consideration of certain terms.

An example of this is the conversion of a normative sentence like article 26 of the Civil Code

Civil Code, Article 26.

“Called … adult, or simply older,
the one who has turned eighteen; and minor,
or simply smaller, the one that has not come to fulfill them. “

in a logical formulation based on IF Statements.

Mayoria_edad = 18
Juan_edad = 19

if Juan_edad> = Major_ age:

   print (“of age”)

else:

   print (“is not of legal age”)

Rule-based logic systems have two basic limitations:

    The combination of variables can be very numerous, which leads to an exponential increase in the number of branches (possible solutions). In other words, if the legal problem involves the combination and interaction of many variables, then the number of solutions can be very high, which makes the programming of each of the branches a very expensive task in time and work, or a series of combinations in excess of numerous that make its programming impossible.
    There are variables whose definition depends on a qualitative appreciation of the authority. In many cases the law appeals to normative concepts that are only defined by the judge attended to the facts of the specific case. Thus, it is sometimes impossible to program ex ante, responses to events whose quantitative or binary assessment depends on the determination made by the authority based on considerations that can not be determined in advance.

Expert systems structured on the basis of data analysis using statistical tools or machine learning.

As a counterpart to rule-based systems, expert systems structured in data analysis do not have a written response menu in advance, but are programmed to mathematically analyze a large database of information that serves as an example, the dataset , and based on that training, create a model that allows to draw conclusions and their multiple probabilities derived from the contrast between a new example and the dataset.

The conversion operation of a normative sentence how article 26 of the Civil Code would look conceptually in this way:

dataset = “public deeds where the date of the
contract, the date of birth of the parties and the mention
that are people of legal age “

Model = “99 percent of the people in
18 years or older that appear in contracts
is also referred to as “of legal age” “

Input = (María = 19 years old)
Output = “99% chance of Maria being” of legal age “”

The systems based on statistical analysis and artificial intelligence have become popular in recent years mainly thanks to the continuous decrease in the cost of calculating and storing data and the propagation of open source algorithms to carry out machine learning tasks.

Within the expert systems based on machine learning we can mention a predictive model of the decisions of the Supreme Court of the United States or some of the contract analysis functions of the KIRA software.

The challenge with the models based on data analysis and machine learning is that due to their technical complexity they require an advanced team for their development. On the other hand, the training of a predictive model requires a large amount of data, which is not always easy to obtain when the information consists of hundreds of thousands of documents and more or less confidential contracts.
 Where are we going with legal expert systems?

Expert systems are not new to the legal world, however, even before the last major financial crisis, they may not have thrived due to the lack of economic incentives for their introduction. Today, the decrease in technological costs and the pressure to have more efficient and less expensive tools to access legal information have strongly promoted its rebirth.

In our opinion, an expert system does not in itself replace the work of legal advice tailored. Expert systems are a new product that allows law firms to profitably capitalize on the knowledge of their lawyers.

    Expert systems are a new product that allows law firms to profitably capitalize on the knowledge of their lawyers.

They allow opening new lines of business and help to focus the work of tailor-made consulting, under billable hours, for those complex cases that generate greater profit margin for the firm. Important studies of foreign lawyers have already adopted this strategy offering products, especially in the area of compliance, which are added to their traditional services and generate new returns. At the same time, it can be very useful for companies to complement the advice of their lawyers with access to expert systems, as this can save significant costs in solving routine and frequent legal inquiries whose response logic is automatable.

     …

Comments and questions are welcome to my email [email protected]

Diego Maldonado Rosas

Subscribe to our newsletter