Skip to content

semantic-systems/coypu-temporal-extraction-service-LEGACY

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Temporal Extraction Models

Coypu

Introduction and setting up

This repository is an implementation of the forked repository based on the master's thesis "Extraction and Classification of Time in Unstructured Data" (2023). It dockerizes the projects and through an API allows the access and usage of these services.

The files for the container are in the build folder. The scripts of both services were merged there. If one needs separate files, the original repository contains them on folder names after each service.

In order to build and raise the container one must simply have Docker installed, position themselves on the root folder of the repository and run the following command:

docker compose up --build -d

This will orchestrate the image and container, download the necessary files along with setting up the API.

Endpoints

All the calls must always be accompanied by the parameter key, the value must be the correct password.

The call to both of these endpoints must be done with the POST standard of HTTP calls. The base URL that the call must be performed to is:

https://temporal-extraction.skynet.coypu.org

For both services, the call expects a raw body in JSON format with the value of sentence, like so:

{"sentence": "I will go get bread tomorrow in the morning."}

The parameter key is expected alongside.

The response of each of the services is different, however, they are both returned in a JSON format and encapsulated in the results key.

MACHAMP

To use the service of MACHAMP, one must add /machamp to the base URL when executing the call. Like so:

https://temporal-extraction.skynet.coypu.org/machamp

The result, using the body shown in the Endpoints section, will have the following format:

{
   "results":[
      {
         "token":"I",
         "value":"O"
      },
      {
         "token":"will",
         "value":"O"
      },
      {
         "token":"go",
         "value":"O"
      },
      {
         "token":"get",
         "value":"O"
      },
      {
         "token":"bread",
         "value":"O"
      },
      {
         "token":"tomorrow",
         "value":"B-DATE"
      },
      {
         "token":"in",
         "value":"O"
      },
      {
         "token":"the",
         "value":"O"
      },
      {
         "token":"morning",
         "value":"O"
      },
      {
         "token":".",
         "value":"O"
      }
   ]
}

UIE

To use the service of UIE, one must add /uie to the base URL when executing the call. Like so:

https://temporal-extraction.skynet.coypu.org/uie

The result, using the body shown in the Endpoints section, will have the following format:

{
   "results":{
      "input_text":"I will go get bread tomorrow in the morning.",
      "record":{
         "entity":{
            "offset":[
               [
                  "date",
                  [
                     5
                  ]
               ],
               [
                  "time",
                  [
                     8
                  ]
               ]
            ],
            "string":[
               [
                  "date",
                  "tomorrow"
               ],
               [
                  "time",
                  "morning"
               ]
            ]
         },
         "event":{
            "offset":[
               
            ],
            "string":[
               
            ]
         },
         "relation":{
            "offset":[
               
            ],
            "string":[
               
            ]
         }
      },
      "seq2seq":"<extra_id_0><extra_id_0> date<extra_id_5> tomorrow<extra_id_1><extra_id_0> time<extra_id_5> morning<extra_id_1><extra_id_1>",
      "tagged_sentence":"I will go get bread <TIMEX3>tomorrow</TIMEX3> in the <TIMEX3>morning</TIMEX3>.",
      "tokens":[
         "I",
         "will",
         "go",
         "get",
         "bread",
         "tomorrow",
         "in",
         "the",
         "morning",
         "."
      ]
   }
}

About

Service creation of the orginal project, only with the MaChamp module.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 93.1%
  • HTML 3.4%
  • Shell 3.0%
  • Other 0.5%