Entity Resolution (ER) is the process that determines who is who and who is related to who within your data. Entity Resolution is different to simple record matching as Senzing ER creates a complete résumé of an entity including all the names you know them by, all the places they have lived you have recorded, the email addresses they have used, etc. Senzing ER also figures out, and remembers how resolved entities are related.
The Senzing ER Workbench is specifically designed to support people and organizations. For ER to be effective, these entities will have names, addresses, phone numbers, identifiers and other attributes. WARNING: If you have name only data or names plus very limited information e.g. city or gender, Senzing ER is not for you.
Senzing ER will be perfectly suited to entity resolve your customers, prospects, employees and vendors. Really any data source containing identity data, including watch lists.
The workbench has the following features:
This occurs within seconds and uses very sophisticated fuzzy matching techniques to ensure we return everything your organization knows about that entity and their relationships.
Lets get started!
Locating Your Data Sources
These are your customers and prospects, possibly even your employees and vendors - any list of records containing identity data, including watch lists. You can find these in your:
- Address Books such as: Microsoft Outlook, Gmail and Lotus Notes
- Customer Relationship Management (CRM) systems such as: Salesforce, ACT, SugarCRM, and Microsoft Dynamics
- Direct Marketing Systems such as: Mailchimp, Constant Contact, Marketo and Zoho
- Web and e-Commerce systems such as: Wordpress, WooCommerce and Stripe
- Accounting and HR systems such as: Quickbooks, Zoho, Wave, ADP, and Oracle
- Spreadsheets and other files: Sometimes identifying data about entities is simply kept in spreadsheets or other files. Such is often the case with prospect lists and watch lists
All these systems have import and export functions allowing you to move their entities from one system to another. Look for the option to export to CSV (Comma Separated Values). The Workbench only reads CSV files.
Check out our list of Plug and Play Data Sources for instructions on how to export their entities. Don’t worry if your system is not listed there, take the following steps:
- Look for a menu option or search the help for the terms ”import and export”
- Perform a web search using an expression such as “how to export <entity type> from <your system>”. Example: “How to extract customers from quickbooks”
- If you have an IT department that built and maintains your system, contact them to extract the master entity data
Sometimes you are given the option to choose what fields to export. You will want to select all the Identity Data available. Look for:
- System ID or account number (this is how you would look them up in the source system)
- Names - Primary name, Synonyms, Nicknames, AKA if any
- Date of birth
- Addresses - Home, Mailing, Alternate, etc
- Phone numbers - Home, Work, Fax, Cell or Mobile
- Email addresses - Possibly websites for a company
- All identifiers - SSN, other National ID, Drivers License, Passport or similar
The more of these fields you provide the better the Workbench will be able to determine if an entity in one data source is the same as in another. In fact, for data sources that only have name data, it will not resolve to any other data sources and is hardly worth loading.
Loading Your Data
Once you have identified a data source and exported its entities to a CSV file, it is time to load the exported CSV into the Workbench. Go to the add a data source card and select your csv file.
If it’s one of the Plug and Play Data Sources, you will likely see the status is immediately set to LOAD NOW because the Workbench has auto-mapped the fields. Simply click on LOAD NOW to begin loading the data source.
If not, you will see the status is set to REVIEW MAPPING
Click on that status to go into the mapping screen. Mapping is the process of annotating the columns in a data source with a set of common terms the Workbench uses to resolve entities. There is additional help under the Mapping Help button on the mapping screen and in the Mapping Assistance article. On the mapping screen additionally add a data source name to describe this data.
Once you have completed the mapping, press the Ready to Load button to return to the data source card where you can actually begin the loading process. Once again by clicking LOAD NOW.
When a data source is loading you will see a spinning circle superimposed on the data source card. Loading can take from minutes to hours depending upon how many records are in the file, the speed of your computer, and other factors that will be discussed in a later article. Mouse over the Loading... status to see how fast it is running and how far through the file it is.
Ideally you should not close the Workbench while data is being loaded, though you can close it if you need to. Loading will restart the next time you go into the Workbench, but you will have to press the RESUME LOADING status to resume.
After completion of a data source loading you'll see LOADED, REVIEW RESULTS. Clicking this will take you to the results review screen.
Reviewing Your Matches
After you load each data source for the first time it is important that you review the matches made. This is the final verification that the data has been mapped correctly and has been successfully loaded. For instance, if you expected matches and didn’t get any, it might simply be that you forgot to map an important field like surname or that the given name field is always empty indicating a problem with the export.
You can review your matches from LOADED, REVIEW RESULTS as described above and from the dashboard screen on the card titled REVIEW. When only one data source is loaded, you will simply see three circles on this card: One for the Duplicates that were found, one for the Possible Duplicates, and one for the Possibly Related records.
Clicking on the hyperlink in each circle will take you to the review of that category of match.
When more data sources are loaded, these circles turn into overlapping circles and you can select which two data sources you want to compare via the data source selectors at the top of the card. As before, clicking on the hyperlinks in the circles will take you to the review screens of the appropriate selection.
If you find something that does not look correct and it is not immediately apparent why, check out the troubleshooting article for common errors that can occur and for details on how to collect and send support information to us.
When reviewing the matches you may see an ECL button. This stands for Entity Centric Learning and it indicates that there are additional records that may have contributed to the match. Simply press the ECL button to show those records.
The review screen by default opens in a condensed view. To expand to view and show additional details click the 'Show all columns' slider at the top of the review section.
Single Subject Searches
You can search for entities from the dashboard screen on the card titled SEARCH and by clicking the magnifying glass in the sidebar.
Search was designed to find all the records for a single entity. It is not a query for a group of people who live in the same city or have the same last name, etc. In fact, you must enter the complete name (given and surname/family name) to find a match on name only. Likewise, you must provide a complete address including street address and either city or postal code to find a match on address.
While you can just search by name, address or phone number, etc, imagine someone calling in asking what information you have on them. To be sure you find them and only them, you would need to be provided with more than merely their name. You might have several people with the same name but all at different addresses or with different birthdays or passport numbers.
Ideally a search for a single entity would contain the following:
- A complete name
- A complete address and/or phone and/or email
- A birth date and/or passport, driving license number or national ID
Once you execute the search you will be presented with a list of the matching records by the match level.
You will only get records in the Matches category If you searched for a complete identity as described above. But we always show you Possible Matches and Possibly Related records as well as name only matches just so you have all the records to review when making a decision about which records belong to the identity you searched for.
From the search screen you can utilize the Print and Export buttons. Using print provides the ability to send to a normal printer or if you have PDF printing (or similar) installed, sending to such a format.
Additionally using export enables you to export the results to a CSV file for use in spreadsheets and other applications.