Images, scanned documents, mislabeled documents and mislabeled images, don't throw it all away. There are relatively straight forward clean-up solutions. It would be a mistake to say the solutions are "easy" but they are certainly doable and perhaps less expensive than you might think.
Can you create a cross-reference table (excel spreadsheet will do nicely) with the bad information on one side and the correct information on the other? One-to-many and many-to-one translations can also be used effectively. A formula can be used to define the cross-reference and a program routine written to implement it. Multi-step translations can also be used along with other functions where logic or corrected data can be injected into the conversion clean-up. Defined patterns often fit some of the errors. That same pattern can be used in reverse (an over simplified statement but it does work) to undo the error.
One mistake we see made often is an attempt to clean-up the data with only one pass at the files. A more practical solution is to make several passes at the data. Each time removing from the working file all the images that have been corrected. The left over file is then much easier to work. Obscure error patterns in the data that were buried in a huge volume of records become easier to spot in a smaller file.
A well written "scrubbing" program will have no trouble doing the clean up. The process will work as far as you want to take it. The law of diminishing returns becomes the governing factor in how many passes you make at the old data file.
There are hidden benefits to a data conversion scrubbing process. For example, it is not uncommon for users of the data to discover things about their data that were previously unknown to them. Some things were probably forgotten over time or something from the "previous team" was not passed along in the transition.
Justification for a project of this nature comes from both a financial consideration and a staff availability consideration. Patience is the primary key to make scrubbing work. Investing staff time on your part and some money to pay a conversion consultant will take care of the rest.
Can you create a cross-reference table (excel spreadsheet will do nicely) with the bad information on one side and the correct information on the other? One-to-many and many-to-one translations can also be used effectively. A formula can be used to define the cross-reference and a program routine written to implement it. Multi-step translations can also be used along with other functions where logic or corrected data can be injected into the conversion clean-up. Defined patterns often fit some of the errors. That same pattern can be used in reverse (an over simplified statement but it does work) to undo the error.
One mistake we see made often is an attempt to clean-up the data with only one pass at the files. A more practical solution is to make several passes at the data. Each time removing from the working file all the images that have been corrected. The left over file is then much easier to work. Obscure error patterns in the data that were buried in a huge volume of records become easier to spot in a smaller file.
A well written "scrubbing" program will have no trouble doing the clean up. The process will work as far as you want to take it. The law of diminishing returns becomes the governing factor in how many passes you make at the old data file.
There are hidden benefits to a data conversion scrubbing process. For example, it is not uncommon for users of the data to discover things about their data that were previously unknown to them. Some things were probably forgotten over time or something from the "previous team" was not passed along in the transition.
Justification for a project of this nature comes from both a financial consideration and a staff availability consideration. Patience is the primary key to make scrubbing work. Investing staff time on your part and some money to pay a conversion consultant will take care of the rest.
KW Norris is an IT professional, consultant and Sales Executive. KW works with medical clinics and medical software vendors to provide the best technology solutions available to improve quality and efficiency in the medical office. If you need a technology solution, KW can help you find it. KW Norris Technology Consultants, Inc. 4125 SW 185th Avenue, Beaverton, OR 97006 503-356-4105 ext 11, 503-939-9223 cell phone, 503-356-4109 fax kw@tech-consultants.com http://www.tech-consultants.com/ http://www.linkedin.com/in/kwnorris Article Source: http://ezinearticles.com/?expert=KW_Norris |
No comments:
Post a Comment