Changelog

Improved accuracy of estimated employee counts for publicly-traded, US companies

by Kris Negulescu

We recently improved the accuracy of the estimate populating the 'nbEmployees' attribute for publicly-traded US organizations in the Knowledge Graph by expanding analysis of data from sec.gov reports. 70%+ of the SEC-10k documents now include nbEmployees for each reporting period.

We also added 'secForms' as an attribute of Organizations. Below is a JSON excerpt from IBM featuring a filing from 2019:

"secForms": [
  {
    "formType": "8-K/A",
    "periodOfReport": {
      "str": "d2019-06-30",
      "precision": 3,
      "timestamp": 1561852800000
    },
    "filingDate": {
      "str": "d2019-09-20",
      "precision": 3,
      "timestamp": 1568937600000
    },
    "documentUrl": "https://www.sec.gov/ix?doc=/Archives/edgar/data/51143/000155837019008675/ibm-20190709x8ka.htm",
    "filingUrl": "https://www.sec.gov/Archives/edgar/data/51143/000155837019008675/0001558370-19-008675-index.htm"
  },
  {
    "formType": "10-Q",
    "periodOfReport": {
      "str": "d2019-06-30",
      "precision": 3,
      "timestamp": 1561852800000
    },
    "filingDate": {
      "str": "d2019-07-30",
      "precision": 3,
      "timestamp": 1564444800000
    },
    "documentUrl": "https://www.sec.gov/ix?doc=/Archives/edgar/data/51143/000155837019006560/ibm-20190630x10q.htm",
    "filingUrl": "https://www.sec.gov/Archives/edgar/data/51143/000155837019006560/0001558370-19-006560-index.htm"
  },
  ...
]

Added "customers" as an attribute of Organizations

by Kris Negulescu

We recently added sources to the Knowledge Graph that provide insights into 'service provider <> customer' relationships. The 'customer' attribute is populated for over 50k companies. Here is an example of a customer of IBM's represented in JSON:

"customers": [
  {
    "summary": "Major airline of the United States",
    "image": "https://kg.diffbot.com/image/api/get?fetch=yes&url=g%3Cj7P0St0DnBJf.x0KwLZrUn.%5B%3CR0Aa4Hh%3B%5Bv738ZqOr7U%3FivHo%3Ef%5B1_l%2F%5B9%40%3BAk%7BAt%3C%60e%5Bt%3DH%2FDk%7DO%28YpOr%29.oCJ",
    "types": [
      "Organization",
      "Corporation"
    ],
    "name": "American Airlines",
    "diffbotUri": "http://diffbot.com/entity/ECm88VZ0TMyq8glAXFBMYaA",
    "targetDiffbotUri": "http://diffbot.com/entity/ECm88VZ0TMyq8glAXFBMYaA",
    "targetDiffbotId": "ECm88VZ0TMyq8glAXFBMYaA",
    "surfaceForm": "American Airlines",
    "type": "Corporation"
  },
  {
    "summary": "Telecommunications company",
    "image": "https://kg.diffbot.com/image/api/get?fetch=yes&url=g%3Cj7P0St0DnBJf.x0KwLZrUn.%5B%3CR0Aa4Hh%3B%5Bv738ZqOr7U%3FDvsR%3EgB%7BYr3izczY2.K%7C%3E",
    "types": [
    "Organization",
    "Corporation"
    ],
    "name": "BT Italia",
    "diffbotUri": "http://diffbot.com/entity/EZK7tKHZCNGCUhTjGW7SaKg",
    "targetDiffbotUri": "http://diffbot.com/entity/EZK7tKHZCNGCUhTjGW7SaKg",
    "targetDiffbotId": "EZK7tKHZCNGCUhTjGW7SaKg",
    "surfaceForm": "BT Italia",
    "type": "Corporation"
  },
  ...
]

We will continue to scale up coverage over time.

List API - NEW!

by Kris Negulescu

Today we've added The List API to the Diffbot API suite of extraction APIs. The List API automatically extracts data from any single web page that contains a primary list of items, such as news index pages, product listings pages, and search engine results pages. It is also embedded under the Diffbot Analyze API which will now attempt to identify and extract pages that match the criteria for a List page. Please refer to the List API documentation for more information.

similarTo 'id1 and ID2'

by Kris Negulescu

We've added the ability to query on look-alikes using two or more Diffbot Org IDs as inputs.

similarTo(id:or(id1, id2, ...)): e.g. 'Organizations similar to Target and Walmart'

type:Organization similarTo(id:or("ExADb18D6MAmunRrlVELe8A", "EOU1WEvHYN6K83Etm91H9fQ"))

Estimated revenue for Private Organizations

by Kris Negulescu

We have introduced a new model to predict revenue for private companies. Almost every company in the KG (over 243M) now has revenue either extracted from the web or estimated with this model.

Try 'similarTo' Searches

by Kris Negulescu

Use the Knowledge Graph to find look-alikes:

type:Organization similarTo(type:Organization name:"Walmart")