arXiv subjects

Daniel Martin Katz

Publications and source records attributed to Daniel Martin Katz.

At least 19 recordsLinked to original sources

arXiv2026

Natural Language Processing in the Legal Domain

We summarize the current state of the field of NLP & Law with a specific focus on recent technical and substantive developments. To support our analysis, we construct and analyze a nearly complete corpus of nearly one thousand NLP & Law related papers published between 2013-2024. Our analysis highlights several major trends. Namely, we document an increasing number of papers written, tasks undertaken, and languages covered over the course of the past decade. We observe an increase in the sophistication of the methods which researchers deployed in this applied context. Legal NLP is beginning to match not only the methodological sophistication of general NLP but also the professional standards of data availability and code reproducibility observed within the broader scientific community. We believe all of these trends bode well for the future of the field and point to an exciting next phase for the Legal NLP community.

cs.CL↗

arXiv2025

GRAPH-GRPO-LEX: Contract Graph Modeling and Reinforcement Learning with Group Relative Policy Optimization

Contracts are complex documents featuring detailed formal structures, explicit and implicit dependencies and rich semantic content. Given these document properties, contract drafting and manual examination of contracts have proven to be both arduous and susceptible to errors. This work aims to simplify and automate the task of contract review and analysis using a novel framework for transforming legal contracts into structured semantic graphs, enabling computational analysis and data-driven insights. We introduce a detailed ontology mapping core legal contract elements to their graph-theoretic equivalents of nodes and edges. We then present a reinforcement learning based Large Language Model (LLM) framework for segmentation and extraction of entities and relationships from contracts. Our method, GRAPH-GRPO-LEX, incorporates both LLMs and reinforcement learning with group relative policy optimization (GRPO). By applying a carefully drafted reward function of graph metrics, we demonstrate the ability to automatically identify direct relationships between clauses, and even uncover hidden dependencies. Our introduction of the gated GRPO approach shows a strong learning signal and can move contract analysis from a linear, manual reading process to an easily visualized graph. This allows for a more dynamic analysis, including building the groundwork for contract linting similar to what is now practiced in software engineering.

cs.AI↗

arXiv2025

The KL3M Data Project: Copyright-Clean Training Resources for Large Language Models

Practically all large language models have been pre-trained on data that is subject to global uncertainty related to copyright infringement and breach of contract. This creates potential risk for users and developers due to this uncertain legal status. The KL3M Data Project directly confronts this critical issue by introducing the largest comprehensive training data pipeline that minimizes risks related to copyright or breach of contract. The foundation of this project is a corpus of over 132 million documents and trillions of tokens spanning 16 different sources that have been verified to meet the strict copyright and licensing protocol detailed herein. We are releasing the entire pipeline, including 1) the source code to acquire and process these documents, 2) the original document formats with associated provenance and metadata, 3) extracted content in a standardized format, 4) pre-tokenized representations of the documents, and 5) various mid- and post-train resources such as question-answer, summarization, conversion, drafting, classification, prediction, and conversational data. All of these resources are freely available to the public on S3, Hugging Face, and GitHub under CC-BY terms. We are committed to continuing this project in furtherance of a more ethical, legal, and sustainable approach to the development and use of AI models.

cs.CL↗

arXiv2025

Precise Legal Sentence Boundary Detection for Retrieval at Scale: NUPunkt and CharBoundary

We present NUPunkt and CharBoundary, two sentence boundary detection libraries optimized for high-precision, high-throughput processing of legal text in large-scale applications such as due diligence, e-discovery, and legal research. These libraries address the critical challenges posed by legal documents containing specialized citations, abbreviations, and complex sentence structures that confound general-purpose sentence boundary detectors. Our experimental evaluation on five diverse legal datasets comprising over 25,000 documents and 197,000 annotated sentence boundaries demonstrates that NUPunkt achieves 91.1% precision while processing 10 million characters per second with modest memory requirements (432 MB). CharBoundary models offer balanced and adjustable precision-recall tradeoffs, with the large model achieving the highest F1 score (0.782) among all tested methods. Notably, NUPunkt provides a 29-32% precision improvement over general-purpose tools while maintaining exceptional throughput, processing multi-million document collections in minutes rather than hours. Both libraries run efficiently on standard CPU hardware without requiring specialized accelerators. NUPunkt is implemented in pure Python with zero external dependencies, while CharBoundary relies only on scikit-learn and optional ONNX runtime integration for optimized performance. Both libraries are available under the MIT license, can be installed via PyPI, and can be interactively tested at https://sentences.aleainstitute.ai/. These libraries address critical precision issues in retrieval-augmented generation systems by preserving coherent legal concepts across sentences, where each percentage improvement in precision yields exponentially greater reductions in context fragmentation, creating cascading benefits throughout retrieval pipelines and significantly enhancing downstream reasoning quality.

cs.CL↗

arXiv2025

KL3M Tokenizers: A Family of Domain-Specific and Character-Level Tokenizers for Legal, Financial, and Preprocessing Applications

We present the KL3M tokenizers, a family of specialized tokenizers for legal, financial, and governmental text. Despite established work on tokenization, specialized tokenizers for professional domains remain understudied. Our paper offers two main contributions to this area. First, we introduce domain-specific BPE tokenizers for legal, financial, and governmental text. Our kl3m-004-128k-cased tokenizer uses 9-17% fewer tokens than GPT-4o and Llama3 for domain-specific documents, despite having a smaller vocabulary. For specialized terminology, our cased tokenizer is even more efficient, using up to 83% fewer tokens for legal terms and 39% fewer tokens for financial terms. Second, we develop character-level BPE tokenizers (4K, 8K, and 16K vocabulary sizes) for text correction tasks like OCR post-processing. These tokenizers keep consistent token boundaries between error-containing and correct text, making it easier for models to learn correction patterns. These tokenizers help professional applications by fitting more text in context windows, reducing computational needs, and preserving the meaning of domain-specific terms. Our analysis shows these efficiency gains directly benefit the processing of long legal and financial documents. We release all tokenizers and code through GitHub and Hugging Face to support further research in specialized tokenization.

cs.CL↗

arXiv2023

LeXFiles and LegalLAMA: Facilitating English Multinational Legal Language Model Development

In this work, we conduct a detailed analysis on the performance of legal-oriented pre-trained language models (PLMs). We examine the interplay between their original objective, acquired knowledge, and legal language understanding capacities which we define as the upstream, probing, and downstream performance, respectively. We consider not only the models' size but also the pre-training corpora used as important dimensions in our study. To this end, we release a multinational English legal corpus (LeXFiles) and a legal knowledge probing benchmark (LegalLAMA) to facilitate training and detailed analysis of legal-oriented PLMs. We release two new legal PLMs trained on LeXFiles and evaluate them alongside others on LegalLAMA and LexGLUE. We find that probing performance strongly correlates with upstream performance in related legal topics. On the other hand, downstream performance is mainly driven by the model's size and prior legal knowledge which can be estimated by upstream and probing performance. Based on these findings, we can conclude that both dimensions are important for those seeking the development of domain-specific PLMs.

cs.CL↗

arXiv2023

GPT as Knowledge Worker: A Zero-Shot Evaluation of (AI)CPA Capabilities

The global economy is increasingly dependent on knowledge workers to meet the needs of public and private organizations. While there is no single definition of knowledge work, organizations and industry groups still attempt to measure individuals' capability to engage in it. The most comprehensive assessment of capability readiness for professional knowledge workers is the Uniform CPA Examination developed by the American Institute of Certified Public Accountants (AICPA). In this paper, we experimentally evaluate OpenAI's `text-davinci-003` and prior versions of GPT on both a sample Regulation (REG) exam and an assessment of over 200 multiple-choice questions based on the AICPA Blueprints for legal, financial, accounting, technology, and ethical tasks. First, we find that `text-davinci-003` achieves a correct rate of 14.4% on a sample REG exam section, significantly underperforming human capabilities on quantitative reasoning in zero-shot prompts. Second, `text-davinci-003` appears to be approaching human-level performance on the Remembering & Understanding and Application skill levels in the Exam absent calculation. For best prompt and parameters, the model answers 57.6% of questions correctly, significantly better than the 25% guessing rate, and its top two answers are correct 82.1% of the time, indicating strong non-entailment. Finally, we find that recent generations of GPT-3 demonstrate material improvements on this assessment, rising from 30% for `text-davinci-001` to 57% for `text-davinci-003`. These findings strongly suggest that large language models have the potential to transform the quality and efficiency of future knowledge work.

cs.CL↗

arXiv2022

GPT Takes the Bar Exam

Nearly all jurisdictions in the United States require a professional license exam, commonly referred to as "the Bar Exam," as a precondition for law practice. To even sit for the exam, most jurisdictions require that an applicant completes at least seven years of post-secondary education, including three years at an accredited law school. In addition, most test-takers also undergo weeks to months of further, exam-specific preparation. Despite this significant investment of time and capital, approximately one in five test-takers still score under the rate required to pass the exam on their first try. In the face of a complex task that requires such depth of knowledge, what, then, should we expect of the state of the art in "AI?" In this research, we document our experimental evaluation of the performance of OpenAI's `text-davinci-003` model, often-referred to as GPT-3.5, on the multistate multiple choice (MBE) section of the exam. While we find no benefit in fine-tuning over GPT-3.5's zero-shot performance at the scale of our training data, we do find that hyperparameter optimization and prompt engineering positively impacted GPT-3.5's zero-shot performance. For best prompt and parameters, GPT-3.5 achieves a headline correct rate of 50.3% on a complete NCBE MBE practice exam, significantly in excess of the 25% baseline guessing rate, and performs at a passing rate for both Evidence and Torts. GPT-3.5's ranking of responses is also highly-correlated with correctness; its top two and top three choices are correct 71% and 88% of the time, respectively, indicating very strong non-entailment performance. While our ability to interpret these results is limited by nascent scientific understanding of LLMs and the proprietary nature of GPT, we believe that these results strongly suggest that an LLM will pass the MBE component of the Bar Exam in the near future.

cs.CL↗

arXiv2022

LexGLUE: A Benchmark Dataset for Legal Language Understanding in English

Laws and their interpretations, legal arguments and agreements\ are typically expressed in writing, leading to the production of vast corpora of legal text. Their analysis, which is at the center of legal practice, becomes increasingly elaborate as these collections grow in size. Natural language understanding (NLU) technologies can be a valuable tool to support legal practitioners in these endeavors. Their usefulness, however, largely depends on whether current state-of-the-art models can generalize across various tasks in the legal domain. To answer this currently open question, we introduce the Legal General Language Understanding Evaluation (LexGLUE) benchmark, a collection of datasets for evaluating model performance across a diverse set of legal NLU tasks in a standardized way. We also provide an evaluation and analysis of several generic and legal-oriented models demonstrating that the latter consistently offer performance improvements across multiple tasks.

cs.CL↗

arXiv2021

Law Smells: Defining and Detecting Problematic Patterns in Legal Drafting

Building on the computer science concept of code smells, we initiate the study of law smells, i.e., patterns in legal texts that pose threats to the comprehensibility and maintainability of the law. With five intuitive law smells as running examples - namely, duplicated phrase, long element, large reference tree, ambiguous syntax, and natural language obsession -, we develop a comprehensive law smell taxonomy. This taxonomy classifies law smells by when they can be detected, which aspects of law they relate to, and how they can be discovered. We introduce text-based and graph-based methods to identify instances of law smells, confirming their utility in practice using the United States Code as a test case. Our work demonstrates how ideas from software engineering can be leveraged to assess and improve the quality of legal code, thus drawing attention to an understudied area in the intersection of law and computer science and highlighting the potential of computational legal drafting.

cs.IR↗

arXiv2021

Measuring Law Over Time: A Network Analytical Framework with an Application to Statutes and Regulations in the United States and Germany

How do complex social systems evolve in the modern world? This question lies at the heart of social physics, and network analysis has proven critical in providing answers to it. In recent years, network analysis has also been used to gain a quantitative understanding of law as a complex adaptive system, but most research has focused on legal documents of a single type, and there exists no unified framework for quantitative legal document analysis using network analytical tools. Against this background, we present a comprehensive framework for analyzing legal documents as multi-dimensional, dynamic document networks. We demonstrate the utility of this framework by applying it to an original dataset of statutes and regulations from two different countries, the United States and Germany, spanning more than twenty years (1998-2019). Our framework provides tools for assessing the size and connectivity of the legal system as viewed through the lens of specific document collections as well as for tracking the evolution of individual legal documents over time. Implementing the framework for our dataset, we find that at the federal level, the United States legal system is increasingly dominated by regulations, whereas the German legal system remains governed by statutes. This holds regardless of whether we measure the systems at the macro, the meso, or the micro level.

cs.SI↗

arXiv2020

Complex Societies and the Growth of the Law

While a large number of informal factors influence how people interact, modern societies rely upon law as a primary mechanism to formally control human behaviour. How legal rules impact societal development depends on the interplay between two types of actors: the people who create the rules and the people to which the rules potentially apply. We hypothesise that an increasingly diverse and interconnected society might create increasingly diverse and interconnected rules, and assert that legal networks provide a useful lens through which to observe the interaction between law and society. To evaluate these propositions, we present a novel and generalizable model of statutory materials as multidimensional, time-evolving document networks. Applying this model to the federal legislation of the United States and Germany, we find impressive expansion in the size and complexity of laws over the past two and a half decades. We investigate the sources of this development using methods from network science and natural language processing. To allow for cross-country comparisons over time, we algorithmically reorganise the legislative materials of the United States and Germany into cluster families that reflect legal topics. This reorganisation reveals that the main driver behind the growth of the law in both jurisdictions is the expansion of the welfare state, backed by an expansion of the tax state.

cs.SI↗

arXiv2018

OpenEDGAR: Open Source Software for SEC EDGAR Analysis

OpenEDGAR is an open source Python framework designed to rapidly construct research databases based on the Electronic Data Gathering, Analysis, and Retrieval (EDGAR) system operated by the US Securities and Exchange Commission (SEC). OpenEDGAR is built on the Django application framework, supports distributed compute across one or more servers, and includes functionality to (i) retrieve and parse index and filing data from EDGAR, (ii) build tables for key metadata like form type and filer, (iii) retrieve, parse, and update CIK to ticker and industry mappings, (iv) extract content and metadata from filing documents, and (v) search filing document contents. OpenEDGAR is designed for use in both academic research and industrial applications, and is distributed under MIT License at https://github.com/LexPredict/openedgar.

cs.CL↗

arXiv2018

LexNLP: Natural language processing and information extraction for legal and regulatory texts

LexNLP is an open source Python package focused on natural language processing and machine learning for legal and regulatory text. The package includes functionality to (i) segment documents, (ii) identify key text such as titles and section headings, (iii) extract over eighteen types of structured information like distances and dates, (iv) extract named entities such as companies and geopolitical entities, (v) transform text into features for model training, and (vi) build unsupervised and supervised models such as word embedding or tagging models. LexNLP includes pre-trained models based on thousands of unit tests drawn from real documents available from the SEC EDGAR database as well as various judicial and regulatory proceedings. LexNLP is designed for use in both academic research and industrial applications, and is distributed at https://github.com/LexPredict/lexpredict-lexnlp.

cs.CL↗

arXiv2017

Crowdsourcing accurately and robustly predicts Supreme Court decisions

Scholars have increasingly investigated "crowdsourcing" as an alternative to expert-based judgment or purely data-driven approaches to predicting the future. Under certain conditions, scholars have found that crowdsourcing can outperform these other approaches. However, despite interest in the topic and a series of successful use cases, relatively few studies have applied empirical model thinking to evaluate the accuracy and robustness of crowdsourcing in real-world contexts. In this paper, we offer three novel contributions. First, we explore a dataset of over 600,000 predictions from over 7,000 participants in a multi-year tournament to predict the decisions of the Supreme Court of the United States. Second, we develop a comprehensive crowd construction framework that allows for the formal description and application of crowdsourcing to real-world data. Third, we apply this framework to our data to construct more than 275,000 crowd models. We find that in out-of-sample historical simulations, crowdsourcing robustly outperforms the commonly-accepted null model, yielding the highest-known performance for this context at 80.8% case level accuracy. To our knowledge, this dataset and analysis represent one of the largest explorations of recurring human prediction to date, and our results provide additional empirical support for the use of crowdsourcing as a prediction method.

physics.soc-ph↗

arXiv2017

Law on the Market? Abnormal Stock Returns and Supreme Court Decision-Making

What happens when the Supreme Court of the United States decides a case impacting one or more publicly-traded firms? While many have observed anecdotal evidence linking decisions or oral arguments to abnormal stock returns, few have rigorously or systematically investigated the behavior of equities around Supreme Court actions. In this research, we present the first comprehensive, longitudinal study on the topic, spanning over 15 years and hundreds of cases and firms. Using both intra- and interday data around decisions and oral arguments, we evaluate the frequency and magnitude of statistically-significant abnormal return events after Supreme Court action. On a per-term basis, we find 5.3 cases and 7.8 stocks that exhibit abnormal returns after decision. In total, across the cases we examined, we find 79 out of the 211 cases (37%) exhibit an average abnormal return of 4.4% over a two-session window with an average $|t|$-statistic of 2.9. Finally, we observe that abnormal returns following Supreme Court decisions materialize over the span of hours and days, not minutes, yielding strong implications for market efficiency in this context. While we cannot causally separate substantive legal impact from mere revision of beliefs, we do find strong evidence that there is indeed a "law on the market" effect as measured by the frequency of abnormal return events, and that these abnormal returns are not immediately incorporated into prices.

physics.soc-ph↗

arXiv2017

A General Approach for Predicting the Behavior of the Supreme Court of the United States

Building on developments in machine learning and prior work in the science of judicial prediction, we construct a model designed to predict the behavior of the Supreme Court of the United States in a generalized, out-of-sample context. To do so, we develop a time evolving random forest classifier which leverages some unique feature engineering to predict more than 240,000 justice votes and 28,000 cases outcomes over nearly two centuries (1816-2015). Using only data available prior to decision, our model outperforms null (baseline) models at both the justice and case level under both parametric and non-parametric tests. Over nearly two centuries, we achieve 70.2% accuracy at the case outcome level and 71.9% at the justice vote level. More recently, over the past century, we outperform an in-sample optimized null model by nearly 5%. Our performance is consistent with, and improves on the general level of prediction demonstrated by prior work; however, our model is distinctive because it can be applied out-of-sample to the entire past and future of the Court, not a single term. Our results represent an important advance for the science of quantitative legal prediction and portend a range of other potential applications.

physics.soc-ph↗

arXiv2017

Measuring the temperature and diversity of the U.S. regulatory ecosystem

Over the last 23 years, the U.S. Securities and Exchange Commission has required over 34,000 companies to file over 165,000 annual reports. These reports, the so-called "Form 10-Ks," contain a characterization of a company's financial performance and its risks, including the regulatory environment in which a company operates. In this paper, we analyze over 4.5 million references to U.S. Federal Acts and Agencies contained within these reports to build a mean-field measurement of temperature and diversity in this regulatory ecosystem, where companies are organisms inhabiting the regulatory environment. While individuals across the political, economic, and academic world frequently refer to trends in this regulatory ecosystem, far less attention has been paid to supporting such claims with large-scale, longitudinal data. In this paper, we document an increase in the regulatory energy per filing, i.e., a warming "temperature." We also find that the diversity of the regulatory ecosystem has been increasing over the past two decades, as measured by the dimensionality of the regulatory space and distance between the "regulatory bitstrings" of companies. These findings support the claim that regulatory activity and complexity are increasing, and this measurement framework contributes an important step towards improving academic and policy discussions around legal complexity and regulation.

physics.soc-ph↗