Matchmerize
The web-based Data Matching Tool for cases where Artificial Intelligence (AI) just isn't smart enough
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 dataMatchmerize was designed for challenging data matching situations in which spreadsheets and automation tools fail
- The matching difficulty
- The amount of data to match
Besides: In case that there exists a Strong AI, matching data might be humanity's smallest issue ;-)
Matchmerize
Great for situations with high matching difficulty. Keeping the consistency of matching like an automation tool while outperforming spreadsheets regarding speed easily
Using Matchmerize follows 5 steps when working alone or 7 steps with a team
All alone
With a team
No team at hand? No problem!
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.
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
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
Matchmerize is free to use for smaller projects
Free Ticket
0
-
3 Team mates
-
3 Files per project
-
100 Rows per file
Custom-tailored
?
-
Whatever your needs are...
-
Our ambition is to make it possible!
-
The sky is the limit