Tag parts of speech

The tokens that constitute the chat message are tagged with part-of-speech labels such as word classes and punctuation marks. For example, a token may be a word and that word may be a verb. This allows the chatbot to better understand the context of the tokens and the structure of the message.

Zoo Chatbot uses the Penn Treebank part-of-speech tags. For example, a plural noun is denoted by NNS.

Example request

Send the following tokenized message:

POST /zoo-chatbot/tag-parts-of-speech

{
  "tokens": [
    "any",
    "giraffes",
    "?"
  ]
}

Response

Notice that any is identified as a determiner and giraffes is identified as a plural noun. The full stop denotes the end of a sentence:

Response

{
  "tags": [
    "DT",
    "NNS",
    "."
  ],
  "probabilities": [
    {
      "tag": "DT",
      "probability": 0.9201411555
    },
    {
      "tag": "NNS",
      "probability": 0.867740459
    },
    {
      "tag": ".",
      "probability": 0.982848614
    }
  ]
}