What corporate data do humans process?
March 5, 2010 1:53 PM Subscribe
What types of data do companies need humans to process?
What kinds of data are companies processing that require a human touch? Some of this would probably be the types of things Mechanical Turk does, but feel free to think beyond that.
What kinds of data are companies processing that require a human touch? Some of this would probably be the types of things Mechanical Turk does, but feel free to think beyond that.
I had a job processing old land titles. They were scanned and we went through the images and copied the information. It couldn't be done via OCR because much of it was handwriting, some going back a hundred years, and also there was a level of interpretation involved where only the relevant information was copied plus it was put into set formats. It was a pretty big one off task that took a couple of years and everything is automated now, but there have been/will be similar big digitising tasks involving human input. This was called data capture instead of data entry, which may be a term for google or whatever.
I also manually enter a lot of my experimental data into a computer from written worksheets because a piece of paper is often easier and more appropriate recording device than a computer or otherwise. I type fast enough that it's more efficient to just type it in rather than set up any kind of automated system. Then the data gets manipulated and interrogated using excel (pivot tables rule) and genstat and the like, all of which involves heavy user input rather than just running a script. I'm doing this as a student now but we all did similar tasks for a company when I was employed. Most experimental data for biological research needs the human touch at some point to check it's reasonable and work out if and how it's biologically relevant. There are companies dealing with even very large datasets (genotyping and the like) which will involve highly trained researchers working with the data in a number of different ways beyond just running the analysis scripts.
Lastly I've had a job processing student loans and allowances. At the start of the academic year thousands of applications come piling in over a shortish space of time and need to be processed. back then that involved lots of paper applications which we had to open, collate and track whereas much of it is automated now (online applications etc), but there will still be a stage where someone looks at every application and checks that it makes sense and that the form has been signed correctly etc. I assume other loans processing is the same, someone in the bank physically looks at the account records and application so make sure it's how it should be (although how much actual input they have into the decision making process these days I don't know).
posted by shelleycat at 2:12 PM on March 5, 2010
I also manually enter a lot of my experimental data into a computer from written worksheets because a piece of paper is often easier and more appropriate recording device than a computer or otherwise. I type fast enough that it's more efficient to just type it in rather than set up any kind of automated system. Then the data gets manipulated and interrogated using excel (pivot tables rule) and genstat and the like, all of which involves heavy user input rather than just running a script. I'm doing this as a student now but we all did similar tasks for a company when I was employed. Most experimental data for biological research needs the human touch at some point to check it's reasonable and work out if and how it's biologically relevant. There are companies dealing with even very large datasets (genotyping and the like) which will involve highly trained researchers working with the data in a number of different ways beyond just running the analysis scripts.
Lastly I've had a job processing student loans and allowances. At the start of the academic year thousands of applications come piling in over a shortish space of time and need to be processed. back then that involved lots of paper applications which we had to open, collate and track whereas much of it is automated now (online applications etc), but there will still be a stage where someone looks at every application and checks that it makes sense and that the form has been signed correctly etc. I assume other loans processing is the same, someone in the bank physically looks at the account records and application so make sure it's how it should be (although how much actual input they have into the decision making process these days I don't know).
posted by shelleycat at 2:12 PM on March 5, 2010
Mail-In-Rebate companies.
posted by blue_beetle at 3:09 PM on March 5, 2010
posted by blue_beetle at 3:09 PM on March 5, 2010
I've had temp work entering data from census forms into the gov't DB.
posted by Billegible at 3:34 PM on March 5, 2010
posted by Billegible at 3:34 PM on March 5, 2010
The US Postal Service has people working at a handful of Remote Encoding Centers around the country who enter address information from envelopes that computers are unable to read - although the amount of work is shrinking, of course, as text recognition improves.
posted by flod logic at 8:37 PM on March 5, 2010
posted by flod logic at 8:37 PM on March 5, 2010
The IRS still employs live people every year to enter physically mailed-in tax forms. There are fewer of these since so many people file electronically, but still.
It's not exactly data-entry, but does medical transcription fit your question? That has to be done by a human.
posted by Violet Hour at 1:07 AM on March 6, 2010
It's not exactly data-entry, but does medical transcription fit your question? That has to be done by a human.
posted by Violet Hour at 1:07 AM on March 6, 2010
This thread is closed to new comments.
posted by banannafish at 2:00 PM on March 5, 2010