File size: 1,399 Bytes
aa5686f
 
 
 
 
 
 
dba7b60
 
 
047b597
8645612
047b597
68e7638
047b597
68e7638
 
 
aa5686f
 
a20aabc
aa5686f
a20aabc
aa5686f
438d740
 
 
 
 
 
 
 
aa5686f
5e69cf5
aa5686f
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: twitter-health-users
  results: []
widget:
  - text: >-
      We are the #UnitedNations’ health agency - #HealthForAll
  - text: >-
      Journal of Anesthesiology and Pain Therapy provides insight into original research and highlights the latest advancements in anesthesiology
  - text: >-
      Human First. EMDR Therapist | Field Instructor | Dog & Plant Mom
  - text: >-
      Board-certified #Dermatologist from @Harvard
  - text: >-
      Human. Person. Father. Huge Real Madrid fan
---

Use this model to detect Twitter users' profiles related to healthcare. User profile classification may be useful when searching for health information on Twitter. For a certain health topic, tweets from physicians or organizations (e.g. ```Board-certified dermatologist```) may be more reliable than undefined or vague profiles (e.g. ```Human. Person. Father```).

The model expects the user's ```description``` text field (see [Twitter API](https://developer.twitter.com/en/docs/twitter-api/v1/data-dictionary/object-model/user) docs) as input and returns a label for each profile:

- `not-health-related`
- `health-related`
  - `health-related/person`
  - `health-related/organization`
  - `health-related/publishing`
  - `health-related/physician`
  - `health-related/news`
  - `health-related/academic`

F1 score is 0.9