Patented Technology

Technology

The following diagram from Patent 6,236,993 will help clarify our unique, patented technique as it automatically discovers two files (1) and (2) with different dates are identical (except for the dates). Note the following:

It is unnecessary to specify the starting location or format of any specific mismatching token (e.g. dates) within the files. It is also unnecessary to specify the difference between the tokens (e.g. the time period that the file was aged for the second file).

The process works even if the format of the token changes between runs. Speed and accuracy can be improved by limiting the list of what token formats may be found in the file.

The process is completely automatic by only specifying token formats that may occur in the file. Our utility contains a parser that identifies hundreds of thousands of different formats, and more are added upon demand.

The process digitally verifies that mistmatching tokens are the expected ones. It was facinating to watch programmers checking for date problems (during Y2K) when the first step was throw away the date information and look for secondary effects (like calculated interest for 100 years rather than 1 month). We said back then, if you are looking for problems in dates, "don't throw out the primary source of information". Now we say, "if you are not interested in the differences in the time units, recognize those elements, then verify the rest of the files are identical".

Our utility controls which format sets to check for in the files by switches such as one that instructs the utility to "check for dates with 3-digit years." Other switches inhibit default behavior (such as "do not check for 2-digit years").

The shaded area are portions of the files that match identically.

This utility uses sound engineering processes, not silver bullet magic. It utilizes the computer's memory to compute the difference between the two files. The description of the process is as follows:

First, the files are compared until a mismatching character is detected between the two files.

The first mismatch is detected at the (3) symbol. Here, every possible interpretation of dates that include the mismatching byte are interpreted and added to a list. Assuming only 2 and 4 digit years are present in the file and all dates are within the 20th and 21st century. The possible dates from file (1) are:

Feb 20, 1901 (ymd format)
Jan 2, 1920 (mdy format)
Feb 1, 1920 (dmy format)
Jan 2, 2000 (mdy format)
Feb 1, 2000 (dmy format)
Feb 20, 2001 (ymd format)
Jan 2, 2020 (mdy format)
Feb 1, 2020 (dmy format).

Every possible date is also computed for the second file. Here, the dates can be:

Jan 2, 1998 (mdy format)
Feb 1, 1998 (dmy format)
Jan 2. 2098 (mdy format)
Feb 1, 2098 (dmy format).

The difference between every possible date is computed. Since there are 8 dates possible for the first file and 4 dates possible for the second file, there are a total of 32 differences computed. Each of these differences is placed on the list of potential differences for the file. The only time that differences are added to the list of potential differences for the file is during this first step.

The comparison process continues in the files until the next difference is detected at (4). Again, every possible date that includes the mismatching bytes are computed and all possible differences between those dates are computed. The only possible date for the "1999365" is Dec 31, 1999 - a Lilian date format. There is also only one date possible for the "12/31/97." So, the difference between the dates in the file is 2 years and the diagram reflects the possible interpretations of the dates. Any previously discovered potential difference that does not occur in the current list are eliminated from the list of potential difference for the files.

The process ends when there are no more differences in the list of potential differences ("the files are different") or the end of the files are reached ("the files match"). In this specific illustration, only the dates left are valid since there is a difference of exactly 2 years between the files.

This is a description of one illustration from the patent. It does not describe the whole patent which generally covers all applications of this technique to compare files (whenever there could be some ambiguity on the interpretation of a date, time, currency, temperature, etc. contained within the files.) To view the full patent, use the Patent Office Patent Number Search database and search for Patent No. 6,236,993.