Step 3 - Setting Match Group Rules

Guide contents:

• Match Group Rules
• Transforms
• Match Key

Note: This process is the same for both Single-Entity Matching and Across-Entity Matching

This section details how to modify and standardise your data via predefined transforms. This data will then be utilised for matching purposes. Within this section you can modify custom transformations at a session level.

After clicking on 'Match Group Transforms' under 'Settings', you will navigate to the screen below:

Rules: On the left-hand side, a pre-defined list of 'Match Rules' will appear in alphabetical order (ascending-by default). You can drag and drop transform(s) to configure against any Match Group(s). After you drag and drop a transform, a pop-up will appear for you to configure the parameters used within the transform.

Test Input Phrase: This section lets you test your applied transforms. Each line will show you the corresponding transform output.

In this screenshot, only one transform has been selected which only provides one extra output.

Note: You must input your test phrase on the same line as the group folder.

For a detailed overview of our data transformations, please see our Match Group Transforms Guide.

Match Group Rules

During the matching process – data transformations can alter the way that DQ for Dynamics looks at your data (it does not change the actual data). This is very effective for transforming specific data elements purely for the purpose of matching. 

For example, in the 'company name' field you may have a record of TrueData Ltd and another record with TrueData Plc. In this case the “business element” (Ltd or Plc) may be considered irrelevant so you only wish to match on the core of the word “TrueData”. You would use transformations to “exclude” the business elements in this case. See individual transformation categories for a more detailed explanation.

NOTE: Users cannot apply transform rules on the multi-select option set.

There are a variety of Match Group Transforms available:

You can configure the transform parameters for a "Custom Exclude to allow you to work with delimiters stored within the database. Simply insert the left and right delimiter and select your mode. 

This will search for a customised word or phrase and replace it with a custom string. This can be used when standard transforms(trim, upper case, remove characters, etc) may not be flexible for a specific business rule.

This is a flexible feature which allows you to bespoke/tailor a match to cater for the unique nature of each user's data. You can configure transform parameters for ‘Custom Transform Library’, choosing from the 'Category' dropdown list of custom transformations already loaded into the App or you can create and choose from your own custom transformations once added. 

NOTE: Custom Transforms can be configured by navigating to the ‘Custom Transform Library Configuration’ screens in 'Advanced Settings' option under the 'Settings' menu.

This is used to extract a set number of characters or sections from the left or right of the data based on the position, length, pattern or character type. 

This is used to eliminate whole word(s) from left or right end of the data, based on its position.

This is used to eliminate all Vowels, Consonants, Numbers, Punctuation and Other Characters from the data.

Remove Diacritics

This transform will allow you to divide a single text value string into multiple parts based on a custom delimiter, position or separator. You also have to option to return the result with or without the specified delimiter.

This is used to Normalise, Exclude, Elaborate or Abbreviate standard elements from your data. Allowing you to uniform key words within the transform parameters.

Trim String

This will eliminate any trailing spaces at the beginning and the end of a string. There is no screen to configure the transform parameters for ‘Trim String’. You can directly drag and drop the ‘Trim String’ rule for any group.

For a more detail and working examples of each, please see our Match Group Transforms Guide.

Once the data has been transformed for the purpose of matching, it can then be tokenised using a fuzzy matching algorithm. This can be applied by selecting the ‘Match Key’ drop-down. The 'Match Key' is used to select an algorithm for phonetic match token generation - all have their uses, however, we recommend using our DQFonetix algorithm to give the best results. 

The 'Match Key' drop-down will have six choices:

Soundex retains the first letter of the input string to formulate its match token. Soundex removes vowels (a, e, i, o, u) and h and w from the input string. The remaining letters are assigned numbers using a lookup table to produce a token of 4 characters.

This means ‘Cathy’ and ‘Kathy’ will not match as their match tokens begin with a ‘C’ from Cathy and a ‘K’ from Kathy. As such, Soundex does not match well where the start of a word sounds the same but is not the same. Also, due to the numeric substitution it is possible to be shown non-matches (false positive) matches.

DQSoundex overloads Soundex with the advanced capabilities of DQFonetix™. This improves the start of word logic and modifies the first letter(s) of an input string. DQSoundex will de-pluralise and pre-process the start of words to manage variances like ‘C’ to ‘K’ as in 'Cathy' and 'Kathy', as well as ‘Ph’ as in Phonetix to ‘F’ in Fonetix.

Metaphone improves the Soundex algorithm by using information about variations and inconsistencies in English spelling and pronunciation, to produce a more accurate encoding.

This allows you to find more precise matches than the simple Soundex algorithm. Metaphone considers a larger set of character transformations than Soundex and therefore analyses a string phonetically with far more accuracy.

DQMetaphone like DQSoundex is an enhanced Metaphone technology with the advanced capabilities of DQFonetix™. This improves the start of word logic and modifies the first letter(s) of an input string. DQSoundex will de-pluralise and pre-process the start of words to manage variances and improve matching.

In the case shown below (Christopher), Metaphone would have generated three of five names matches. However, after running DQ’s advanced algorithms and advanced logic, DQMetaphone allows ‘Kh’ from 'Khristopher' to match with the ‘Ch’ from 'Christopher'. Thus generating the same match key token.

DQFonetix™ contains our advanced phonetic algorithms developed by DQ Global over the last 25 years. The algorithm is property of DQ Global and hence we do not share the specification of the process. However, DQFonetix™ has four key features:

• Five spoken languages – English, Spanish, French, Italian and German
• Avoids false matches
• Overcomes character variances
• Deals with diacritics

DQFonetix™ provides your CRM system with the most varied matching window allowing you to highlight duplicate matches that may not be picked up – or falsely matched - in Soundex and Metaphone.

Selecting no match key will not generate a phonetic token, hence no match token will be generated. However, this allows you to match identical strings.