AI Integrations Documentation

Overview

This subsystem is designed to provide suggestions based on machine learning models.

Installation

Notice! It is recommended to use this library starting from version 1.1.0 that introduces authentication via an API key.

Node

Install with npm:

npm install prozorro_ai

Browser

Include the latest script directly from npm.

<script src="https://cdn.jsdelivr.net/npm/prozorro_ai@1.1.0/prozorro_ai.min.js"></script>

OR

  1. Download the latest distribution in releases.
  2. Make sure to serve it from your webserver.
  3. Include it on the client from a SCRIPT tag.

Usage

Import

Import script to use Prozorro_AI.

For webpack:

import Prozorro_AI from "prozorro_ai"

For requirejs:

const Prozorro_AI = require('prozorro_ai')

For cdn:

<script src="https://cdn.jsdelivr.net/npm/prozorro_ai@1.1.0/prozorro_ai.min.js"></script>

Authentication

This library requires authentication via an API key. The API key is issued after the registration confirmation on this website. You can find it on the Account page.

Use API key as follows:

Prozorro_AI.client({apiKey: 'Insert your API_KEY here'}).units.suggest({
...
})

Item Unit suggestion

Based on the inputs, you can get the most probable unit of measurement.

Try demo
property description type required default
1 tenderTitle A name of the tender string required at least one of № 1-4 ' '
2 tenderDescription A description of the tender string required at least one of № 1-4 ' '
3 itemDescription A description of the goods, services to be provided string required at least one of № 1-4 ' '
4 itemClassification The primary classification for the item. It is mandatory for classification.scheme to be CPV. string required at least one of № 1-4 ' '
5 accuracyCutoff Probability threshold string/number not required 0.1
6 limit Number of the results to be displayed string/number not required 5

The example would be as follows:

Find unit of measurement based on tender's title and description, item's description and classification.

Request / response example:

Item Classification suggestion

Based on the inputs, you can get the most probable classification.

Try demo
property description type required default
1 tenderTitle A name of the tender string required at least one of № 1-4 ' '
2 tenderDescription A description of the tender string required at least one of № 1-4 ' '
3 itemDescription A description of the goods, services to be provided string required at least one of № 1-4 ' '
4 itemUnit Name of the unit based on the UN/CEFACT Recommendation 20 unit code. string required at least one of № 1-4 ' '
5 accuracyCutoff Probability threshold string/number not required 0.1
6 limit Number of the results to be displayed string/number not required 5

Find classification based on tender's title and description, item's description and unit of measure.

Request / response example:

Examples

Various examples can be viewed here