HTML 5 support, e.g. Safari, Chrome, Firefox, Edge, IE 11
Data Analytics
Data Matching
Windows, OSX, Linux
)

Matchmerize Logo Matchmerize
The web-based Data Matching Tool for cases where Artificial Intelligence (AI) just isn't smart enough

The web-based Data Matching Tool for cases where Artificial Intelligence (AI) just isn't smart enough
A simple matching task
solved easily with text-similarity based fuzzy matching
Matching a list of US President names to a comparable list works well using text-similarity. However, there are weaknesses, e.g. the false matching between L. Johnson and A. Johnson.
Find out what makes sense here!
A hard matching task
where a deeper understanding of the context is required
Matching the US President names to nicknames is much harder. Real knowledge is required. This is what a human being can do easily, in particular with a supporting tool.
Learn how this works with Matchmerize!

Data Matching - A key preparation step in most data science projects

In many data scientists' dreams it would be possible to concentrate on analyses, insights, predictions and all the other things making fun.

In real life there is lots of data preparation required upfront, e.g. data cleansing. Often, when combining different data sources, data matching (aka data harmonization, aka data linkage) has to be done as well. It is required to connect entries between unrelated data sets for overarching analyses.

Often this can be done programmatically. When a common identifier exists it is really easy, when text-similarity or structures can be used it might work as well. In all other cases, Matchmerize allows it to leverage human capabilities efficiently.

Learn more about the shortcomings of text-based similarity when matching data

Matchmerize was designed for challenging data matching situations in which spreadsheets and automation tools fail

For data matching various tools exists. However, not all tools are suited for all situations. Basically, there are two aspects to consider for the decision:
  1. The matching difficulty
  2. The amount of data to match
In all cases data matching requires human effort, either for matching or for validating algorithm-based matchings.

Besides: In case that there exists a Strong AI, matching data might be humanity's smallest issue ;-)

Spreadsheets

Great for smaller amounts of data, in particular when VLookup works. With multiple lists to handle, high chance of contradictory matchings. Slow and error-prone. E.g. Excel, Numbers

More about spreadsheet complications

Automation Tools

Great for large amounts of data when text-similarity helps. Often, manual validation of matchings required. For difficult matchings useless. E.g. Knime, R, Alteryx

More about limits of text-similarity

Matchmerize

Great for situations with high matching difficulty. Keeping the consistency of matching like an automation tool while outperforming spreadsheets regarding speed easily

More about how Matchmerize works
More about spreadsheet complications
More about limits of text-similarity
More about how Matchmerize works

Using Matchmerize follows 5 steps when working alone or 7 steps with a team

All alone

The first step is to create a new Matchmerize project. All you need is a project name.
The second step is about uploading the data sets you want to match. For each data set you just need to select the column which holds the entries for matching. At this time CSV and Excel files are supported.
The third step is about making settings. Most importantly, you need to decide how the data sets shall be matched, e.g. List A vs List B or maybe a combination of List A and List B against themselves.
The fourth step is about matching the data yourself. In case you need support, you can invite team mates any time. If you dont have anyone ready, we can provide a commercial offer for a team.
Finally, all your matchings will be consolidated into one truth. This contradiction-free matching list can then be downloaded as Excel file.

With a team

The first step is to create a new Matchmerize project. All you need is a project name.
The second step is about uploading the data sets you want to match. For each data set you just need to select the column which holds the entries for matching. At this time CSV and Excel files are supported.
The third step is about making settings. Most importantly, you need to decide how the data sets shall be matched, e.g. List A vs List B or maybe a combination of List A and List B against themselves.
The fourth step is getting your supporters work for you. You provide their email addresses and Matchmerize sends them login information via email. Dont worry, if your team has different opinions. Matchmerize creates a consistent output in the end.
The fifth step is about letting the team match the data. You can jump in and support this, of course.
The sixth step is for judging whether the team did it correctly. So all matching pairs are listed and can be down-voted if you see that one is not correct.
Finally, all matchings of the team mates and your judgement (optional) will be consolidated into one truth. This contradiction-free matching list can then be downloaded as Excel file.

No team at hand? No problem!

We can give you a hand and provide the brain power your matching task requires. Feel free to reach out!

Matchmerize is different than existing solutions for data matching

Matching list entries

Matchmerize is for matching lists. Each entry has some text (e.g. product names) and maybe an identifier (e.g. product numbers).

One or more data sources

Matchmerize can help to match entries from a list against itself (which is like duplicate finding) or between multiple lists.

When AI is not smart enough

Matchmerize is meant for difficult tasks where text-based matching makes no sense and such algorithms fail.

Human wisdom makes the difference

Matchmerize enables humans to effectively leverage their cognitive power solving difficult matching tasks.

Ready, steady, go!

Matchmerize is quickly ready for matching. Uploading CSV or Excel files, selecting the columns and matching can start.

Faster than expected

Matchmerize has a UX designed for matching data quickly. It outperforms techniques using text editors and spreadsheet easily.

The team is the star

Matchmerize allows multiple people to work on a project. Combined human brain power delivers even better results.

Perfect like a diamond

Matchmerize fosters matchings carefully, avoiding conflicts and contradictions. Particularly, when multiple people match data.

Start the Matchmerize workbench tour

Use Case #1: Price comparisons between product lists obtained through online-shop web-scraping

Situation

A web scraping tool collected hundreds of product names and prices. A price comparison shall be done. It requires matching the products based on their names.

Complication

Differences in the product names are hard to identify programmatically. This leads to lots of false positives (matchings which do not belong together), e.g. "Phone A, Red, 4 core CPU, ..., 128GB, ..." and "Phone A, Red, 4 core CPU, ..., 256GB, ..." will be easily mistaken for the same as there is only a 3 character differences.

Question

If programmatic matching is so hard, can a person solve this task? And how can this person be supported to become highly efficient?

Watch this video and learn how it works

Video coming soon

Use Case #2: Pharma data analysis involving not easily linkable 3rd-party data sources

Situation

For a pharmaceutical data science project various API performance indicators need to be calculated. The involved data sets come from numerous origins e.g. business warehouse, market research, web scraping data, etc.

Complication

Fortunately, (fuzzy) text matching on API names works pretty well. However, the data sources contain chemical names (e.g. ascorbic acid), trivial names (e.g. vitamin C), formulas (e.g. C6H8O6), translations ("Ascorbinsäure"), E numbers (E 300). Consequently, programmatically matching these data sources will be rather complicated.

Question

If programmatic matching is so hard, can a person solve this task? And how can this person be supported to become highly efficient?

Watch this video and learn how it works

Video coming soon

Matchmerize is free to use for smaller projects

Free Ticket

0

  • 3 Team mates

  • 3 Files per project

  • 100 Rows per file

Sign up for free No credit card required

Custom-tailored

?

  • Whatever your needs are...

  • Our ambition is to make it possible!

  • The sky is the limit

Get a quote