Services API - Sotera/watchman GitHub Wiki

Watchman Services Endpoints

4 min screencast

Watchman is a data analytics toolkit that allows rapid prototyping through a series of REST endpoints that analyze both text and images. Endpoints can be programmatically chained together to create small, powerful apps, without the burden of installing 3rd party software, handling deployments, or writing custom code. All you need is an HTTP client within any programming language, or just use cURL to try it out.

API root: /api

Content-type: application/json

POST /translate/detect

# Request
{"text": "حاصرت القوات الكردية، المشاركة في الهجوم على تنظيم الدولة الإسلامية لاستعادة مدينة الموصل العراقية، بلدة بعشيقة، شمالي البلاد."}

# Response
"ar"

POST /translate/en

text to English

# Request
{"text": "حاصرت القوات الكردية، المشاركة في الهجوم على تنظيم الدولة الإسلامية لاستعادة مدينة الموصل العراقية، بلدة بعشيقة، شمالي البلاد."}

# Response
"The Kurdish forces surrounded, participating in the attack on the Islamic State to reclaim the northern Iraqi city of Mosul, a town with a mistress, the North of the country."

POST /extract/textract

# Request
{"text": "<div><div>Mosul Iraq</div></div>",
"mime_type": "text/html"}

# Response
"Mosul Iraq"

POST /extract/entities

# Request
{"text": "<p>The Kurdish forces surrounded, participating in the attack on the Islamic State to reclaim the northern Iraqi city of Mosul</p>",
"mime_type": "text/html",
"extract_type": "mitie"} # mitie, stanford

# Response
[
  {
    "score": 1.0291898759148814,
    "tag": "ORGANIZATION",
    "label": "Islamic State"
  },
  {
    "score": 1.1319111014626635,
    "tag": "MISC",
    "label": "Iraqi"
  },
  {
    "score": 1.4876167809342686,
    "tag": "LOCATION",
    "label": "Mosul"
  }
]

POST /extract/addimageurls

Note: used in conjunction with /getcaptions for neuraltalk2 processing

Request
["http://ichef.bbci.co.uk/news/ws/736/amz/worldservice/live/assets/images/2016/10/24/161024073836_calais_camp_640x360_reuters_nocredit.jpg"]

Response
# returns an array of objects with 'sha256sum' property. pass those values to /getcaptions

POST /extract/getcaptions

Request 
# sha id's from /addimageurls response
["734f4"]

Response
[
  {
    "caption": "a group of people standing next to a train"
  }
]

POST /geocoder/forward-geo

Request
{"address": "Mosul Iraq"}

Response
[
  {
    "latitude": 36.34585235,
    "longitude": 43.098576921383,
    "country": "العراق",
    "city": "Mosul",
    "state": "نینوى",
    "countryCode": "IQ"
  },
  {
    "latitude": 36.3466059,
    "longitude": 43.127008,
    "country": "العراق",
    "city": "موصل",
    "state": "نینوى",
    "countryCode": "IQ"
  }
]

POST /extract/features

Note: Used in conjunction with /jobs/status endpoint. Currently, must be a public access .zip file.

Request
{"archive_url": "https://s3.amazon.com/path/to/file.zip"}

Response
[{"filename": "file1.jpg", "features": [1, 0.5, 2, ...]}, ...]
# returns job_id for querying jobs/status endpoint

POST /jobs/status

{"job_id": "abc123"}

# returns object with 'status' and 'data' properties. data is a feature vector.
⚠️ **GitHub.com Fallback** ⚠️