Path: newssvr20.news.prodigy.com!newsmst01.news.prodigy.com!prodigy.com!in.100proofnews.com!in.100proofnews.com!news-out.visi.com!petbe.visi.com!rcn!feed3.news.rcn.net!not-for-mail Date: Thu, 11 Dec 2003 23:07:39 -0500 From: Jerry Gitomer Reply-To: jgitomer@erols.com User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225 X-Accept-Language: en-us, en MIME-Version: 1.0 Newsgroups: comp.databases.theory Subject: Re: A Question on Integrety References: In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Lines: 47 Message-ID: <3fd93f22$0$14972$61fed72c@news.rcn.com> NNTP-Posting-Host: 208.59.96.237 X-Trace: 1071202083 reader3.news.rcn.net 14972 208.59.96.237:33001 X-Complaints-To: abuse@rcn.com Xref: newssvr20.news.prodigy.com comp.databases.theory:22945 Tobin Harris wrote: > Hi there, > > I have a client who's business is to collect and improve data, wich is > eventually published in a telephone directory. Initially, the data is > allowed to be in a poor state: certain job roles require only finding > minimal data and throwing it in. Later down the line, someone else has to > improve the quality of this data so that it can be printed in the phone > directory, perhaps by adding additional information or refining that already > there. There are also many other stages where the rules governing the data > are contextually sensitive to the current stage of the data in it's > lifecycle. > > For me, this raises an interesting question about data integrety. How do I > design a database for entities that are subject to different business rules > throughout their lifecycle!? I could easily set up the relational integrety > for any one "stage", but this would not be applicable at all times. For > example, if data is ready for print, it has to conform to many rules. > However, if it is undergoing quality control, then there are far less rules. > > I ask this mainly becuase it's not something I've come across when learnign > about logical or physical modelling, and I was just wondering if anyone else > had!? > > Cheers, > > Tobes > > > Some possible approaches to your problem: 1. Put raw data in files and don't move it into the database until it is fully validated. 2. Have two tables, one for validated data and one for unvalidated data. Move after validation complete. 3. Add a bit mask column where each bit represents one of the columns requiring validation. As a column is validated in a row set the appropriate bit in the bit mask. This approach would allow you to extract fully validated rows for inclusion in your directory and to generate reports, statistical or detailed, on any selected validation criteria. HTH