We have been asked a few times now for a more detailed explanation of how the Match weightings work and how they affect the Match score.
The purpose of the Match weighting is to allow you to determine the importance that you want to place on each data description.
The match score is calculated by adding together the score of all the data descriptions and dividing by the number of descriptions. The weighting determines what percentage of the score for each data description is applied to the match score.
i.e. (The example below will show how the match weighting affects the match score:)
- Company Name – Weighting 100%
- Address – Weighting 100%
- City – Weighting 50%
- Postcode – Weighting 100%
Each description adds to the match score as below:
- Company Name contribution to MatchScore = (100% / 350% = approx 29%)
- Address contribution to MatchScore = (100% / 350% = approx 29%)
- City contribution to MatchScore = (50% / 350% = approx 14%)
- Postcode contribution to MatchScore = (100% / 350% = approx 29%)
Therefore by adjusting the match weighting you can determine the importance and contribution that each data description adds to the match score and can therefore adjust these weightings to help find new matches emphasising different elements of your data.
We have been asked on several occasions why you need to specify which country a list pertains to and why can’t we just have an international setting.
The simple answer is accuracy, but let’s not leave it at that, I’d like to share with you some of the nuances that we see on a regular basis to help you understand a little of the work that has gone on behind the scenes, to ensure you get the best matching results possible.
In the coming months I will look to discuss country specific nuances in deeper detail and share the learnings and experiences that we have had matching data for large multi-national organisations across the globe. And show some of the approaches we have taken to tackle these issues at Match2Lists.
Needless to say, it’s not a case of one logic rule works for all, different elements of an address need to be processed and interpreted in different ways and country nuances can and do have significant impact on these rules. By specifying which country your list is related to helps Match2Lists use the most appropriate rules and logic to achieve the best matching results for you.
Several people have asked on occasion if it was possible to increase the width of the Match Results, so one of the other amendments we have made to the visualizer is the addition of sizing options for validating your match results.
Using these buttons you can increase the size of the windows displaying the match results making it easier to read longer company names and addresses. A useful feature when looking for the best possible match.
You can also switch back to the smaller option to view more of potential matches side by side without scrolling – this is particularly helpful when your de-duping a list or looking for multiple matches.
Today we launched an updated version of our acclaimed Match Visualizer which includes new functionaly to continue our quest to speed up the process of manual verification of Match Results.
The new changes include a filter option alongside the Match Weightings that enable you to filter out less likely match results and chunk up the approvals to an even greater degree than ever before.
The Automatic approval process works in conjunction with these new filters, which enables you to verfiy and automatically approve even more matches and oonce again radically cut down the time required for complex data matching like never before.
We will be adding a new video soon to show how you can make the most effective use of this new feature.
Yesterday I had a discussion with one of our Beta Testers who was confused what to do when their reference dataset had two company name fields and they were unsure which one should they match against. I thought it might be a good idea to address this here. In short if you have multiple company names you should be using them during the matching process, as either of them could exist in the list you are matching to.
Read More Post a comment (0)