The sequence similarity formula originated to satisfy this amazing criteria:

• A genuine expression of lexical similarity – strings with tiny variations should always be named getting comparable. In particular, an important sub-string overlap should point out a high standard of similarity between your strings.
• A robustness to changes of word order- two chain that incorporate equivalent statement, however in another purchase, must seen as being close. In contrast, if one string merely a random anagram associated with figures contained in the some other, it should (usually) getting named dissimilar.
• Words independence – the formula should work not only in English, but in several languages.

Answer

The similarity are computed in three strategies:

• Partition each sequence into a listing of tokens.
• Processing the similarity between tokens through the use of a string edit-distance algorithm (extension ability: semantic similarity description by using the WordNet Fort Collins escort library).
• Processing the similarity between two token lists.

There is certainly another conversation for the research.

A much better similarity positioning formula for varying duration strings

Many thanks all for the support and ideas.

Martin Xie [MSFT] MSDN society service | suggestions to you Get or demand rule trial from Microsoft Kindly make every effort to mark the responds as answers when they let and unmark all of them if they give no services.

• Marked as address by Martin_Xie Monday, September 26, 2011 8:48 are

All replies

Understanding your query,explain it considerably more particular,i have mistaken for our

Including “a_logfile.txt” and “logfile_a.txt” need really similiar and aswell “loga_file.txt” and “logfile.text” but not “myText.txt” and “logfile.txt”

Whether it resolved your problem,Please click “level while Solution” on that post and “tag as Helpful”. Delighted Programming!

Alright we try it again 🙂

Better I do want to contrast filenames and I also need a share amounts in how similiar they have been. We dont know if that is feasible at all.

As an example a filename “a_filename.txt” and “filename_a.txt” is really similiar for people but how should I get the exact same outcome programmatically.

Another instance filename “file_abc_.txt” and fil_abc_e.txt” can similiar but yet again how do I have the lead programmaticaly

That is probably tougher than it appears initially.

Have a look at http://en.wikipedia.org/wiki/String_metrics and adhere many backlinks.

Regards David Roentgen Every regimen ultimately turns out to be rococo, and then rubble. – Alan Perlis the sole valid measurement of signal high quality: WTFs/minute.

Welcome to MSDN Community Forum.

This short article demonstrates a great choice about: how-to calculate the similarity between two words/strings. The formula was developed in C# and download the demo inside.

The string similarity algorithm was created to meet the following criteria:

• A genuine reflection of lexical similarity – strings with tiny variations should always be thought to be being close. In particular, a significant sub-string convergence should point out a high level of similarity involving the chain.
• A robustness to adjustment of term purchase- two chain which contain exactly the same keywords, in a new order, need thought to be are comparable. Conversely, if an individual sequence is a random anagram of figures included in the different, it should (usually) feel seen as dissimilar.
• Code freedom – the algorithm should work not only in English, but additionally in many different languages.

Answer

The similarity was computed in three actions:

• Partition each string into a summary of tokens.
• Processing the similarity between tokens simply by using a sequence edit-distance algorithm (expansion feature: semantic similarity dimension using the WordNet collection).
• Computing the similarity between two token records.

There is another conversation for the guide.

A significantly better similarity score algorithm for adjustable size strings

Thanks a lot all to suit your assist and ideas.

Martin Xie [MSFT] MSDN Community help | comments to united states bring or consult laws test from Microsoft Please take the time to draw the replies as responses as long as they let and unmark them as long as they render no services.

• Marked as solution by Martin_Xie Monday, September 26, 2011 8:48 have always been

We have written a rule for my job to discover similar names more or less from database.

first I made use of the DIFFERENCE(string1, string2)>=4 purpose of SQL servers nevertheless did not help me because for example whenever first name is “21” and next label got “21 jump street” the end result contained two names whereas demonstrably they didn’t even comparable. therefore, the lead pair of these a query contained over 700 prices that was very poor in such a case.

however found a similar CHANGE function for c# that has been nearly just like SQL type of that purpose. like it coordinated the similarity of “asdcdfsdfgdsgdg” and “asdewwetqwetrwe” as Perfect that is certainly incorrect.

then I developed a category because of this issue to obtain additional efficient similarity between strings.

title of the class was StringCompare and here is an introduction to this class:

WHAT IS STRING EVALUATE?

StringCompare is a contrasting instrument for strings. Not an ordinal evaluation, but a relative evaluation that establishes how much cash two strings include similar or just how much perhaps not close.

By establishing the great tradeoff prices you may get a beneficial assessment for strings.

WAYS TO USE:

First you need to create an instance of StringCompare with tradeoff prices or default tradeoff principles.

You’ll find 4 beliefs that can be set:

1. MinSimilarityLong:

Here is the minimal acceptable amount of similarity between two chain that evaluating with StringCompare. This advantages is employed for strings together with the length of no less than 8.

2. MinSimilarityShort:

This is actually the minimum appropriate percentage of similarity between two chain that researching with StringCompare. This importance is utilized for chain using length below 8.

3. MaxToleranceLong:

This is basically the optimum acceptable percentage of threshold between two strings that researching with StringCompare. This appreciate is used for strings with the length of at least 8.

4. MaxToleranceShort:

Here is the max acceptable percentage of threshold between two chain that contrasting with StringCompare. This benefits is employed for strings making use of length below 8.

* After you’ve produced an example it is possible to contact InstanceName.IsEqual (string1, string2) to ascertain the equality of two strings.

* think about your equality is actually in accordance with the minSimilarty and maxTolerance you arranged before.

* think about that greater minSimilarity values will result in more limited outcome and vice versa.

* give consideration to that reduced maxTolerance principles can lead to much more restricted results and vice versa.

As An Example: