Legal practice can gain much from advances in machine learning and A.I. technology. Common law countries such as the United States, the United Kingdom, Canada, and Australia, rely on judicial precedent to decide what law governs new cases and how that law is applied. Judicial precedent is the body of case law that observes how past courts have decided similar cases. Case law is composed of individual case decisions by courts that resolve a particular dispute. When resolving new cases, courts look back on judicial precedent to structure how to decide said new cases. Thus, an extensive knowledge base of how prior cases have been decided under what law is needed in each field of legal practice. Legal research is the practice of identifying relevant case law for a particular set of facts and is essential for any practitioner seeking to make the most persuasive argument for clients. Reviewing case law and identifying relevant cases can be time consuming and imperfect; there is always risk that practitioners overlook relevant and/or new case law that has changed the legal landscape of their practice. Thus practitioners are not only required to retain an extensive knowledge base of relevant case law, but also to keep abreast of changes in case law.

This research project develops a methodology to utilize machine-learning analysis of live disputes to assist legal professionals in narrowing issues and comparing relevant precedents. Our first step is to extract and classify sentences in breach of contract court decisions according to type. Such court decisions have five basic sentence types: sentences on contract law, sentences on contract holding, sentences on contract issues, sentences on contract reasoning, and sentences on contract facts. The result of this project facilitates further downstream processing such as constructing decision trees that predict the likely outcome for the case at hand, displaying the rationales on which court decisions are based, and calculating the similarity of previous legal precedents.

In this paper, we present two types of XML semantic constraints, namely the XML Template Functional Dependencies (XTFDs) and the XML Template Multivalued Dependencies (XTMVDs), which are both defined over the finite XML files that have no mixed contents. Essentially they specify that if a pattern of nodes appears in a finite XML file that has no mixed contents, then another pattern of nodes must also appear. To implement these semantic constraints, we demonstrate an algorithm that transforms an XTFD or an XTMVD to an XQuery program such that the program returns a non-empty set of counter examples from such an XML file If [Formula: see text] if and only if [Formula: see text] does not satisfy the constraint.

XML has too many low-level details that hinder high-level conceptual design. We therefore propose DTD graphs and XSD graphs as a mean for conceptual modeling of XML applications. Similar to document type definitions (DTDs) and XML schema definitions (XSDs), DTD graphs and XSD graphs are trees, and as such they can easily be mapped to DTDs and XSDs for implementation. Unlike DTDs and XSDs, DTD graphs and XSD graphs capture various high-level data semantics such as cardinality, ISA, participation, aggregation, categorization, and n-ary relationship. Furthermore, this paper also presents transformation rules between DTDs and a large class of XML Schemas and an algorithm that inputs users’ requirements and outputs a DTD graph that has a minimum number of redundancy-free fragments. As a result of these good properties, the resulting DTD or XSD facilitates query processing and update.

Given an acyclic database scheme R and a connected subtree T of a join tree for R, this paper provides a characterization of the set of multivalued dependencies (MVDs) that hold over T with respect to infinR, the corresponding join dependency (JD) of R. This result corrects an error in one of my previous papers "A Comparative Study of Various Nested Normal Forms," which was published in vol. 14, no. 2, pp. 369-385, Mar./Apr. 2002.

As XML data becomes more and more prevalent and as larger quantities of data find their way into XML documents, the need for quality XML data organization only increase. One standard way of structuring data well is to reduce and, if possible, eliminate redundancy, while at the same time making the storage structures as compact as possible. In this paper, we present a methodology to generate XML storage structures where conforming XML documents are redundancy-free, and for most practical cases, are also fully compact. Our methodology assumes the input is a conceptual-model hypergraph. For the special case that every edge in the hypergraph is binary, we present a simple algorithm, guaranteed to always generate redundancy-free storage structures. We show, however, that generating a minimum number of redundancy-free storage structures is NP-hard. We therefore provide heuristics to guide the process and observe that these heuristics result in satisfactory solutions, which are often optimal. We then present a general algorithm for n-ary edges and show that it generates redundancy-free storage structures. The general algorithm must overcome several problems that do not arise in the special case

In this paper, we model business workflows using Harel’s statecharts. Mapping to statecharts allows us to systematically identify potential workflow problems. It also allows us to investigate specific properties inherent in actual business workflows. Our research focuses on three desirable properties of active database systems — termination, confluence, and observable determinism. For termination and confluence, we develop algorithms to provide a theoretical lens linking desirable active database system properties to workflow management systems problems. We initially validate our algorithms by mapping business workflows from a case study. Our research thus builds preliminary theory by developing a systematic method for identifying workflow problems.

As object-relational databases (ORDBs) become popular in the industry, it is important for database designers to produce database schemes with good properties in these new kinds of databases. One distinguishing feature of an ORDB is that its tables may not be in first normal form. Hence, ORDBs may contain nested relations along with other collection types. To help the design process of an ORDB, several normal forms for nested relations have recently been defined, and some of them are called nested normal forms. In this paper, we investigate four nested normal forms, which are NNF [20], NNF [21], NNF [23], and NNF [25], with respect to generalizing 4NF and BCNF, reducing redundant data values, and design flexibility. Another major contribution of this paper is that we provide an improved algorithm that generates nested relation schemes in NNF [20] from an a-acyclic database scheme, which is the most general type of acyclic database schemes. After presenting the algorithm for NNF [20], the algorithms of all of the four nested normal forms and the nested database schemes that they generate are compared. We discovered that when the given set of MVDs is not conflict-free, NNF [20] is inferior to the other three nested normal forms in reducing redundant data values. However, in all of the other cases considered in this paper, NNF [20] is at least as good as all of the other three nested normal forms.

We give a straightforward definition for redundancy in individual nested relations and define a new normal form that precisely characterizes redundancy for nested relations. We base our definition of redundancy on an arbitrary set of functional and multivalued dependencies, and show that our definition of nested normal form generalizes standard relational normalization theory. In addition, we give a condition that can prevent an unwanted structural anomaly in nested relations, namely, embedded nested relations with at most one tuple. Like other normal forms, our nested normal form can serve as a guide for database design.