Reliable legal decision support, through the use of predictive analytics, could enable legal professionals to build better arguments.

While researching for more valuable information and knowledge on predictive analytics, I came across an exceptionally well-crafted and researched thesis, written by Celia Pienaar in 2020, in which the author effortlessly articulates with historic connotations, the profound use and need for predictive analytics in the modern world.

''When preparing for litigious matters, legal professionals would analyse the matter at hand and search for similar cases supporting their client’s case as well as their opponent’s, to understand the aspects that could impact a judge’s decision-making process. The same process would be undertaken by the judge (or the court assistants) in considering all possible outcomes as part of the decision-making process. The challenges raised with IE on legal texts (not necessarily limited to case law) include that legal texts are usually long (the sentence structure as well as actual document length) and complicated. Most cases cite previously decided cases either in support or distinction, which creates complicated citation networks. While it does not happen often, case judgements could be criticised as decided incorrectly or in line with outdated legislation, or even reversed at a later stage. One noted advantage on the nature of case judgements?is that texts tend to be well written and error free due to a high level of care and precision taken in producing them (Tran et al., 2013). In summary, while case judgements might be highly structured documents as far as the formalities and rules around layout, formatting, content and reference style are concerned, the data they contain is generally unstructured. Also, judges do not typically review thousands of cases but probably only a few hundred at most (Lodder and Oskamp, 2010).

2.2 Legal informatics

Legal informatics is defined as “the discipline which deals with the use of ICT [Information and Communications Technology] to process legal information and support legal activities, namely, the creation, the cognition and the application of the law” (Biasotti et al., 2008). Legal professionals’ roles have been described as finding information, analysing information, and deciding based on such analysis. On the first task described; indices and citator databases are available to assist. On the second; case summaries and relevance rankings offer support. On the third; practice guides assist with the decision-making task but the majority of technology focuses on the first task of finding information. Research giant Thomson Reuters suggests that large datasets create patterns from which legal professionals could benefit when identifying correlations between these patterns and outcomes (Conrad and Al-Kofahi, 2017). An early example of legal informatics is the first legal computer application that was developed at the University of Pittsburg in 1956. It was tasked with finding references to “retarded child” in legislation and replacing it with “exceptional child”. Legislation was transferred to magnetic tape for searching purposes, which lead to the first successful execution of electronic legal text retrieval (Biasotti et al., 2008).

In Italy in 1963 a judge at the Court of Cassation created a database of abstracts of his own decisions, later expanded to also include laws and regulations. This database was subsequently developed into a prominent information system still used today. His example was followed by other nations such as France’s Supreme Judge in Administrative cases, Germany’s Minister of Justice, Sweden’s Directorate of Court Administration and Finland’s Supreme Administrative Court; creating a vision of unified storage of legal sources (Biasotti et al., 2008).

2.3 AI in law

The concept of Artificial Intelligence (AI) was officially proposed at the Dartmouth Conference in 1956, meaning research of AI and law is already at least 60 years developed (Zhang et al., 2019). Edwina Rissland’s famous 1990 expression on projects on AI and law still ring true (Rissland, 1990; as cited by Dadgostari et al., 2020): “A unifying theme of the projects is the goal to understand and model legal argument, a keystone of an overarching goal to understand and model legal reasoning. These goals require that we know first how to represent several types of knowledge, such as cases, rules and arguments; second, how to reason with them, such as to manipulate precedents, to apply and make inferences with rules, and to tailor arguments to facts; and third, how to use them ultimately in a computer program that can perform tasks in legal reasoning and argumentation, such as analogizing favorable cases and distinguishing contrary ones, anticipating parries in adversarial argument, and creating artful hypotheticals.” While the concept is not new, until recently there has not yet been much active research on the application of ML to case law analysis and prediction (Katz, 2012).

An example mentioned is the work done by political scientists Martin and Quinn and legal scholars Ruger and Kim during 2003 where three methods of prediction were applied and ultimately, the machine did fare better than the experts in predicting outcomes (Ruger et al., 2004). More recent sources indicate that there has been further research, without comprehensive evaluation models for such methods (Liu and Chen, 2017; Zhang et al., 2019).

2.4 Predictive analytics An area often written about but not yet part of mainstream legal technology available for commercial use is that of predictive analytics of decision-making; to identify and analyse patterns in historic cases to predict (hopefully with some explanations) future outcomes in similar cases. Already in1970, the growing number of court cases was the focus of research into methods of IE to alleviate the burden on legal professionals (Marx, 1970). As far back as 1993 the vast amount of legal material was described as the “crisis of law” (Schweighofer and Winiwarter, 1993). Although judges may not realise or wish to admit it, their approach to a judgement is a routine in that a mental protocol is followed and it is therefore structured, which lends itself to ML application (Remus and Levy, 2017). Outcome prediction is both historical (focusing on similar previous cases) and empirical (with statistical ML focusing on feature extraction to strengthen or weaken classification) (Raghupathi et al., 2018). Predictive analytics in this context refers to the process of identifying similar or relevant case law, analysing such, assessing validity of historic arguments, construing new arguments and predicting outcomes on future matters based on underlying patterns. Part of this process is to reduce the number of potentially responsive documents by finding legally similar documents and then extracting key legal concepts or rules from those. Its purpose is not so much the number-crunching 10 often associated with big data analysis, but rather the identification and subsequent analysis of underlying patterns.

2.4.1 Benefits The application of predictive analytics to judicial decision-making holds various potential benefits: Improved management of case volumes According to a telephone interview with a leading case law publisher in South Africa on 8 October 2019, case volumes are as high as 3,000 per year (not including all courts) with as little as 15% of that at most being reported (in other words, summarised and circulated to the broader legal community as publicly available information). High volumes of cases not only take a long time to process, thereby placing an enormous burden on the judiciary, but also add to legal professionals’ workload in having to stay abreast of legal decisions relevant to their area(s) of practice.

Public interest

Predictive analytics facilitate public transparency on how the law is interpreted and applied to real-life scenarios. In an ideal world, judgement results should be consistent across similar cases so predictive data models could assist in addressing this challenge.

Reducing the costs of legal services.

One explanation behind the high fees associated with legal services could be the expanding volumes of legal information to be digested by legal professionals. The last two decades have seen many large law firms investing in specialist knowledge officers to assist with this task (many of them former lawyers 11 themselves). Smaller firms and sole practitioners do not necessarily have the luxury of dedicated knowledge professionals tasked with identifying and distributing current awareness.

Reliable legal decision support in the form of a computerised knowledge assistant could enable legal professionals to build better arguments in less time (Cardellino et al., 2017), especially if combined with visual data representation interfaces which would allow them to also explain their arguments more effectively to clients (Conrad and Al-Kofahi, 2017). In addition, legal professionals being able to analyse typical trial length and/or award levels associated with particular types of cases could assist in providing certainty around pricing upfront, and thereby better manage client expectations at the onset of a matter.

Predictability Predictive analytics could play an important role in maintaining the rule of law by improving general predictability - providing even the layman with an understanding (albeit basic) of key legal principles. This could facilitate improving access to justice, and perhaps even improving trust in the judicial system. While it is debatable whether this would reduce the number of legal disputes, it should at the very least improve predictability and accuracy in certain types of decisions, which could eventually reduce the burden on courts due to an increase in early settlement.

Training value

Many commercial legal technology platforms used in other areas of practice such as document creation mentioned above play a role in educating and guiding the next generation of legal professionals. This is achieved by embedding guidelines and 12 principles as part of the applications’ processes. One can argue that predictive analytics available in an understandable format could serve this purpose, potentially levelling the playing field between generations and legal professionals from different demographical or educational backgrounds.''

you can read the full thesis at: https://open.uct.ac.za/bitstream/handle/11427/33925/thesis_sci_2021_pienaar%20celia.pdf?sequence=1&isAllowed=y

We at Legal Chain, are fully commensurate with the inherent intricacies associated with predictive analytics and that it cannot be applied to all legal practice areas, therefore we follow a bespoke law analysis ,in order to effectively meet our predictive outcomes.