Thursday, September 11, 1997

CPS-615 Assignment 1

David Figatner

Assignment Instructions: Please spend some time examining work that you are interested in within your own department, out there on the web, or wherever. Pick a particular problem where parallel computing has been applied. Build a web page describing briefly the application, the use of parallelism, and your assessment of its success, weaknesses, and challenges. If it is not known whether or not parallelism has been applied to your problem, then describe why simulation of the problem would be important. Also, if the problem can be parallelized, what would be the advantages of this parallelism, etc.

Complete assignment and parallel computing links: http://osprey4.npac.syr.edu:7777/cps615-fall97/index.html

EXAMINED DEPARTMENT Legal Document Databases (e.g., WestLaw and Lexis-Nexis)

Search, storage and retrieval of data within legal database
Possibility of more indepth searches ("smart" searching)

STATUS OF PARALLELIZATION OF EXAMINED DEPARTMENT Unknown. Although, there has been some work done on parallel searches through large databases, I do not think this has been applied to any legal document databases.

IMPORTANCE OF PROBLEM Legal research is of paramount importance for law firms to successfully represent their clients. The first step in litigation or contract formation is research. Today, research has been mostly computerized onto large databases such as WestLaw and Lexis-Nexis. Although these databases allow quick searches and retrievals, there is always a place for even faster and more advanced searches. One way that this can be done is through the use of parallel computers and algorithms on the document servers and clients. Since time is money, and the more time the computer spends searching the less time the lawyer has to look over the found documents, the use of parallelization for queries into the database could result in better utilization of a law firm's time and money.

PARALLELIZATION OF THE PROBLEM As of now, databases are contained on servers and searched through client machines. Although this is in a small way a parallel setup, it does not utilize true parallelizational algorithms to improve speed. By having client computers which can concurrently conduct a single search, query times could be drastically cut down, and more advanced searches can be conducted on larger and larger sets of documents.

Legal documents (i.e., cases, law journals and statutes) are being continuously added to these legal databases. In order to keep up with this continuous stream of new documents, there is a need for faster computers and faster searches. Currently, computers are still improving in speed and can handle the mounting number of documents. But, when computer engineering slows down in 10 years, there will be a need for more advanced computer systems to handle the ever increasing legal document load. The computer system will have to be designed to exploit parallelization to increase speed and robustness of the search.

DISADVANTAGES OF PARALLELIZATION The advantages of parallelization have already been mentioned above. The disadavantages of using parallel algorithms and computers on this problem is the same as on other simulation and computer problems. There is a lack of successful and easy to implement programming language and hardware capabilities to implement this technology. Furthermore, because the speed of searching is currently so high, legal database companies do not see the future advantage of researching and developing parallel algorithms and computers to improve search time. Until computers stop advancing in speed and functionality, it will be difficult for parallelization to become an important part of the legal and business community.

NOTES The parallelization of legal databases is only an example of how parallel machines/systems could exploit this developing technology. Any type of database can be improved by parallelization (e.g., business databases, financial history databases). Of course, simulation on top of these databases would also be a valid use of this technology. An example of this would be to have a large database containing financial history which is accessed in parallel connecting with a running parallel simulation which does future financial forecasting.