1.3 Comparisons to other kinds of human rights data projects

The phrase "human rights data" means many things, but this paper is only about one kind of project: the representation of acts of violence in a way that allows human rights researchers to make systematic, comparative analyses of patterns of violation in time, space, and social structure. The kinds of organizations that are interested in such a project are groups that analyze patterns of violence. Such groups may be directly monitoring ongoing conditions or studying historical events. These organizations might include grassroots human rights NGOs, U.N. human rights monitoring missions, government human rights bodies, or truth commissions.

Analyzing patterns of violence is not the only kind of human rights work, and so the information management system described in this paper is not the only kind of database that can serve human rights organizations. Other kinds of human rights organizations, such as documentation centers, advocacy groups working on countries other than the one in which they are located, international criminal tribunals, or groups working with legal questions such as asylum or human rights theory, may have less call for an approach to a human rights history that focuses on direct evidence of particular violations. For example, an advocacy group that follows cases happening in countries other than the groups own country might need to keep a database of their cases. However, the basic element in this kind of database would be the individual person whose case is being followed. The advocacy group might represent brief narrative summaries of what has happened to the victim, but they would have no need to represent detailed codes of the actual violations. Such a group might produce a statistical analysis of their own work, but this would be very brief. This hypothetical advocacy group depends for information on fact-finding missions, press reports, or informants in the country of study. Given these sources, their information is unlikely to be comprehensive, i.e., having coverage across all types of violation in their country of interest, or consistent, i.e., having coverage equally across many years or across other neighboring countries. Thus they are more likely to focus their analysis on trends in areas that fall within their own group's focus or mandate, not overall trends in the violent patterns in a given country. Therefore they probably do not need to build a system with the focus described here.

To describe what this kind of large-scale information management project entails, it may first be useful to distinguish it from several other kinds of very valuable but quite distinct kinds of human rights data. One such project is DIANA, a collaborative effort among many legal libraries which has collected documents, principally treaty instruments, relevant to human rights in a set of interlinked World Wide Web pages on the Internet [1]. Similarly, the U.N. High Commission on Refugees has collected a large number of documents relevant to refugee situations on a CD called REFWORLD. In these exemplary text databases, a user can search for particular pieces of text in the bodies of the work, or for snippets of titles. Full text databases provide fast, easy reference to documents, and they can include information from many different sources, about a vast range of topics. For example, such databases might include the text of treaties, newspaper articles from online sources, human rights organizations' reports, and the text of legal opinons. Databases of these kinds are very flexible, easy to use, and relatively easy to put new data into. Text databases need not have a single focus: they can include information about anything, and new documents can be added as the database's owners choose. Definitions about what should be in a full-text database can change over the life of the database, and such databases can provide references to any query the users request -- even though many queries may return no results.

Data in the sense used in this paper is something different: data are the combinations of information and classifying decisions made by an organization that together serve as the organization's collective memory.

As mentioned above, information management systems include the collection of information and the representation of that information according to the structure defined by the organization. Although the information management system is more than just a database, a structured, relational database is at the core of an information man-agement system. Structured, relational databases are very different from free text databases. An example will probably clarify the differ-ence. If, for example, a researcher using a free-text database wants to see every case relating to electric shocks, she can search for text that refers to electric shocks in a variety of ways ("electric blows," "torture by electricity," "the telephone," "shocks," etc.). The researcher searches for words or phrases which may or may not represent the idea of interest. Many documents, including case files, are likely to emerge with this kind of search, and inevitably, important references are missed. A structured system, however, uses a controlled vocabulary so that any violation that is classified as "electric shock," for example, necessarily fits within the organization's definition of this kind of violation. A search can be absolutely specific and yield all of those cases, and only those cases, in which shocks were used. By framing the search differently, the researcher might list all those people who suffered electric shock torture, organized by names or by pro-fession; or she could list all the police stations in which instances of this kind of torture have been alleged. If the organization includes other kinds of documents in their information management system, documents classified as having some relevance to the notion of shocks might also come out of the search, but documents would be identified with a quite different structure from acts of violence. Similarly if a researcher is searching for cases that occurred in a particular time, or at a particular place, a structured database could yield precisely these cases [2].

The ability to generate simple or very complex statistics is another important feature of structured databases, especially in information management systems that include thousands or tens of thousands of individuals' testimony. Perhaps the most fundamental work that a structured database can do that an unstructured database cannot do is count and tabulate things. The organization can create a wide variety of statistics and graphics based on those counts.