Back to LanguageTool Homepage - Privacy - Imprint

Handling of unknown names

We got a bunch of reports for false positives, where names unknown to LT are considered spelling mistakes. Now I tried to reproduce this with the public API and found dfferent results with the sentence “Hey, are you Bliblaaa?”

curl -X POST --header ‘Content-Type: application/x-www-form-urlencoded’ --header ‘Accept: application/json’ -d ‘text=Hey%2C%20are%20you%20Bliblaaa%3F&language=en-US&enabledOnly=false’ ‘https://api.languagetoolplus.com/v2/check

This gives me an empty match list, but if I run 5.7 snapshot locally (same for 5.5. and 5.6), then I get a spelling mistake on “Bliblaaa”

  "matches": [
    {
      "message": "Possible spelling mistake found.",
      "shortMessage": "Spelling mistake",
      "replacements": [
        {
          "value": "Alibaba"
        },
        {
          "value": "BlaBlaCar"
        }
      ],
      "offset": 13,
      "length": 8,
      "context": {
        "text": "Hey, are you Bliblaaa?",
        "offset": 13,
        "length": 8
      },
      "sentence": "Hey, are you Bliblaaa?",
      "type": {
        "typeName": "Other"
      },
      "rule": {
        "id": "MORFOLOGIK_RULE_EN_US",
        "description": "Possible spelling mistake",
        "issueType": "misspelling",
        "category": {
          "id": "TYPOS",
          "name": "Possible Typo"
        }
      },
      "ignoreForIncompleteSentence": false,
      "contextForSureMatch": 0
    }
  ]

There’s an experimental feature running on languagetool.org that avoids some of these false alarms, but the feature is not quite stable yet.

ah ok good to know. is this going to be a premium or open source feature?

I don’t know, we’ll have to get it stable first.