Abstract: “Building upon developments in theoretical and applied machine learning, as well as the efforts of various scholars including Guimera and Sales-Pardo (2011), Ruger et al. (2004), and Martin et al. (2004), we construct a model designed to predict the voting behavior of the Supreme Court of the United States. Using the extremely randomized tree method first proposed in Geurts, et al. (2006), a method similar to the random forest approach developed in Breiman (2001), as well as novel feature engineering, we predict more than sixty years of decisions by the Supreme Court of the United States (1953-2013). Using only data available prior to the date of decision, our model correctly identifies 69.7% of the Court’s overall affirm and reverse decisions and correctly forecasts 70.9% of the votes of individual justices across 7,700 cases and more than 68,000 justice votes. Our performance is consistent with the general level of prediction offered by prior scholars. However, our model is distinctive as it is the first robust, generalized, and fully predictive model of Supreme Court voting behavior offered to date. Our model predicts six decades of behavior of thirty Justices appointed by thirteen Presidents. With a more sound methodological foundation, our results represent a major advance for the science of quantitative legal prediction and portend a range of other potential applications, such as those described in Katz (2013).”
It is a wrap for #ReInventLaw NYC 2014. We finished up with just over 800 folks in attendance for this free, public facing event at the Cooper Union (~725 at the peak of the day according to the security guards who were keeping the count). As the conference co-organizer, I want to thank all of our speakers for speaking, all of our sponsors for sponsoring and all of our attendees for attending!
There are many interesting changes underway within the legal industry. Many of the participants (both speakers and attendees) are part of the innovator / early adopter segment. It was great to connect with everyone. I hope to continue the conversation. More importantly, I look forward to working together to help build the future …
From the Abstract: “This Article proposes a novel and provocative analysis of judicial opinions that are published without indicating individual authorship. Our approach provides an unbiased, quantitative, and computer scientific answer to a problem that has long plagued legal commentators. Our work uses natural language processing to predict authorship of judicial opinions that are unsigned or whose attribution is disputed. Using a dataset of Supreme Court opinions with known authorship, we identify key words and phrases that can, to a high degree of accuracy, predict authorship. Thus, our method makes accessible an important class of cases heretofore inaccessible. For illustrative purposes, we explain our process as applied to the Obamacare decision, in which the authorship of a joint dissent was subject to significant popular speculation. We conclude with a chart predicting the author of every unsigned per curiam opinion during the Roberts Court.” <HT: Josh Blackman>
In partnership with Michigan State University College of Law and Emory Law, today we announce the Beta Pre-Release of a New Web Interface – LegalLanguageExplorer.com. We are just getting started here with this project and anticipate many features that will be rolling out to you in the near future. Please feel free to send us your feedback / comments.
Instant Return of a Time Series Plot for One or More Comma Separated Phrases. The default search is currently interstate commerce, railroad, deed (with plots for each of the term displayed simultaneously).
Feel free to test out ANY phrase of Up to Four Words in length.
Here are just a few of our favorites:
Clear and Present Danger
SCOPE OF COVERAGE:
In the current version, we are offering results for EVERY decision of the United States Supreme Court (1791-2005). We plan to soon expand to other corpora including the U.S. Court of Appeals, etc.
FULL TEXT CASE ACCESS:
Each of the Phrases you search will be highlighted in Blue. If you click on these highlighted phrases you will be taken to the full list of United States Supreme Court decisions that employ this phrase:
Check out the advanced features including normalization and alternative graphing tools.
Daniel Martin Katz, Michael J. Bommarito II, Julie Seaman, Adam Candeub & Eugene Agichtein, Legal N-Grams? A Simple Approach to Track the ‘Evolution’ of Legal Language in Proceedings of Jurix: The 24th International Conference on Legal Knowledge and Information Systems (Vienna 2011) available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1971953
PRESENTATION & HELPFUL TUTORIAL:
Click on the Image Below and You Will Be Directed to our Presentation at 24th International Conference on Legal Knowledge and Information Systems ( Jurix 2011 – Vienna )
This offers some motivation for the project as well as a Brief Slide Based Tutorial Designed to Highlight Various Functions Available on the Site.
Michael J. Bommarito, Building Legal Language Explorer: Interactivity and Drill-Down, noSQL and SQL available at http://www.michaelbommarito.com/blog/2011/12/16/building-legal-language-explorer-interactivity-and-drill-down-nosql-and-sql/
The visualization above is something we are calling the “six degrees” of Marbury v. Madison. It was originally produced for use in our paper Distance Measures for Dynamic Citation Networks. Due to space considerations, we ended up leaving it on the cutting room floor. However, the visual is designed to highlight the idea of a “sink.”
Sinks are one of the core concepts which we outline in Distance Measures for Dynamic Citation Networks, 389 Physica A 4201 (October 1 2010). Looking through the prism of a citation network, sinks are the root to which a given legal concept, academic idea or patent based innovation can be drawn. From each citation in a non-sink node, it is possible to trace the chains of citations back to their root (which we call a sink). In the visualization above, the root or sink node is the famed United States Supreme Court decision Marbury v. Madison. Starting from the center and working out to the edge, the first ring are cases that directly cite Marbury v. Madison. The next ring are cases which cite cases that cite Marbury v. Madison. The next ring are cases which cite cases which cases that cite Marbury v. Madison and so on…
Anyway, one of the major contributions of our Distance Measures for Dynamic Citation Networks paper is that it allows us to use these sinks to create pairwise distance/similarity measure between the ith and jth unit. In this instance, the units in this directed acyclic network are the ith and jth decisions of the United States Supreme Court.
Now, it is important to note cases contain many citations and thus can be oriented relative to many different sinks. So, even if a case can be traced to the Marbury sink – this does not preclude it from being traced to other sinks as well. Also, it is possible to construct a variety of mathematical functions to characterize the sink based distance between units. For instance, the importance of a sink might decay as its shortest path length increases. An alternative measure might weight the importance of each sinks by the number of unique ancestors shared between nodes i and j that are descended from a given sink of interest. Indeed, many ﬁne-grained choices are possible but they require justiﬁcation drawn from the given substantive problem.
As mentioned above, this method has potential applications including tracing the spread of technological innovation in patent citations or the spread of ideas in a set of academic articles. However, given our primary interest surrounds the judicial citations, we are working on the follow up to the “sinks” paper. In this follow up paper, we hope to carry these and other ideas forward into a definitive community detection method for judicial citation networks.
To preview, at least two major dynamics must be considered in any null model for community detection. First, case-to-case citations can help contribute to the fractal nature of legal systems. In other words, we are pretty far from any sort of gaussian null model. However, this is easy enough to confront with an alternative null — some highly skewed distribution (i.e. power law or power law with a cutoff, etc.)
Here is the difficult part — the cross fertilization of legal concepts. This is a time evolving network where ideas are referenced/imported from otherwise unrelated or previously unrelated domains. The examples of cross-fertilization are numerous. One of my personal favorite non-SCOTUS examples is the use of the tort doctrine of “trespass to chattels” in the context of web scraping.
Anyway, we hope to have more to come on the topic of SCOTUS community detection in the weeks and months to come. In the meantime, please check out a Dynamic 3D Hi Definition United States Supreme Court Visualization.
Kudos to Jerry Goldman, the other folks at the Oyez Project as well as the Chicago-Kent College of Law for making this free resource available to the public!
From the description: “OYEZTODAY at IIT Chicago-Kent College of Law offers you the latest information and media on the current business of the Supreme Court of the United States. OYEZTODAY provides: easy-to-grasp abstracts for every case granted review, timely and searchable audio of oral arguments + transcripts, and up-to-date summaries of the Court’s most recent decisions including the Court’s full opinions. You will have access to all this information on your iPhone with the ability to share reactions on Facebook, Twitter, or by email. (Recordings of opinion announcements from the bench will follow when the Court releases these files to the National Archives at the start of the Court’s next Term). Chicago-Kent is proud to provide this free service to enhance the public’s understanding of the Supreme Court and current legal controversies.”
The Sunday New York Times features an article by Adam Liptak assessing the conservatism of Robert Court. The article features some good coverage for some of the leading law and political science scholars who study the United States Supreme Court. Well worth the read!
Jerry Goldman (Northwestern/ Oyez Project) has recently released a great app for those wanting to quickly access SCOTUS case summaries and/or audio recordings from their Iphones. The top 100 constitutional law cases are made available for free–thanks to the good folks at Justia. Those looking for the full 600+ cases can access them for the low price of $4.99.
Here is the complete description – “PocketJustice brings the U.S. Supreme Court down to earth through abstracts of the Court’s constitutional decisions and access to its public sessions. The application includes voting alignments and biographical sketches for all justices. PocketJustice harnesses recordings of the Court’s public proceedings to deliver hundreds of hours of oral arguments and opinion announcements. In many of these cases, PocketJustice provides synchronized, searchable transcripts identifying all speakers. This version offers information and audio for the top 100 constitutional law cases. The complete version ($ 4.99) provides information and audio for all 600+ constitutional law cases in the Supreme Court canon.”
Along with a non-trivial subset of the legal blogosphere, we eagerly await the Supreme Court’s decision in the Bilski case. Perhaps tomorrow will be the day? In the meantime, here are a variety of thoughts on the matter. Bilski Blog / EFF / Now Europe / PatentlyO / GenomicsLawReport / Ipwatchdog / Fenwick&West / InsideCounsel. This one from Greg Laden is also fun.
Still no announcement from the White House but there has been some movement over at Intrade … Click Here for Most Recent Chart [From Intrade.com]