What are 'Skills' ?

Language Skills

A Language Skill is a package of trained NLP models, available via API. Skills accept the text as input in various formats and respond with processed texts and extracted metadata.

Examples:

Use the Summarize Skill to summarize an article or conversation.
Use the Sentiments Skill to extract sentiments (positive or negative) from a chatbot session.
Use the Sales-Insights Skill to extract insights (like needs and features) from sales calls.

Pipelines

Language Skills are the building stones of the Pipeline API. The pipeline API allows invoking and chaining multiple Skills to process your input text with a single API call. Pipelines can be defined by listing the desired Skills.

curl -X POST \
'https://api.oneai.com/api/v0/pipeline' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H 'api-key: <YOUR-API-KEY-HERE>' \
-d '{
    "input": "I feel very bad about hurting Michelle the other day.",
    "steps": [
      {
        "skill": "sentiments"
      },   
      {
        "skill": "emotions"
      }   
    ]
}'

const OneAI = require("oneai");

const oneai = new OneAI("<YOUR-API-KEY-HERE>");
const text = "I feel very bad about hurting Michelle the other day.";
const pipeline = new oneai.Pipeline(
    oneai.skills.sentiments(),
    oneai.skills.emotions(),
);

pipeline.run(text).then(console.log);

import oneai

oneai.api_key = "<YOUR-API-KEY-HERE>"
text = "I feel very bad about hurting Michelle the other day."
pipeline = oneai.Pipeline(
  steps = [
        oneai.skills.Sentiments(),
        oneai.skills.Emotions(),
  ]
)

output = pipeline.run(text)

Generator Skills vs. Analyzer Skills (why the order is important)

Some skills, e.g. Summarize generate new text based on the input. Other skills. e.g. Emotions, simply extract information from the input without altering it.
When a Skill generates new text, the next skill in the pipeline will use the generated text as input. An analyzer skill simply "passes" the input as it is to the next skill.
In this way when running the two Skills Emotions and Summarize, if Emotions is first it will extract emotions from the input text, but if it is second it will extract emotions from the summary.

const OneAI = require("oneai");

const oneai = new OneAI("<YOUR-API-KEY-HERE>");
const text = "I feel very bad about hurting Michelle the other day.";
const pipeline = new oneai.Pipeline(
    oneai.skills.sentiments(), // extracts sentiments from the original input
    oneai.skills.summarize(),  // summarizes the original input
    oneai.skills.sentiments(), // extracts sentiments from the generated summary
);

pipeline.run(text).then(console.log);

Generated Text

When a Skill generates text, you can find it in the response object under output.text

const OneAI = require("oneai");

const oneai = new OneAI("<YOUR-API-KEY-HERE>");
const text = "I feel very bad about hurting Michelle the other day.";
const pipeline = new oneai.Pipeline(
    oneai.skills.summarize(),
);

pipeline.run(text).then(console.log);

Labels

Most Skills return information in the labels array. A label describes the indices of specific words and sentences in the text that the skill extracted.

{
  "type": "emotion",
  "skill": "emotions",
  "name": "happiness",
  "span_text": "I was delighted to wake up and found out we won another two medals during the night.",
  "span": [
    29,
    113
  ],
  "output_spans": [
    {
      "section": 0,
      "start": 29,
      "end": 113
    }
  ]
}

Label Properties

Property Name	Description
type	the type of the label
skill	the skill that generated the label
name	the name of the label
value	the value of the label
data	additional data
span_text	the portion of the input text that the label applies to
span	the indices of `span_text` in the input.
output_spans	the indices of `span_text` in the output, in case the new text was generated.
input_spans	the indices of `span_text` in the input
timestamp	audio indices, in case the origin of the text is audio
timestamp_end	audio indices, in case the origin of the text is audio