Forum Discussion

PaulaMcC's avatar
PaulaMcC
Frequent Visitor
5 years ago
Solved

Error in Azure Sentiment Analysis

I'm currently using Azure Cognitive Services for the first. I've managed to get the Text Analysis to work but am struggling to get the sentiment analysis to work as it's just returning errors for all rows. 

I'm following the instructions from the MS Tutorial - https://docs.microsoft.com/en-us/azure/cognitive-services/text-analytics/tutorials/tutorial-power-bi-key-phrases plus some additional research & watching video's but I can't work out where I've gone wrong!

 

This is the code I'm using:

let
apikey = "MY.API.KEY",
endpoint = "https://MYSUBDOMAIN.cognitiveservices.azure.com" & "/text/analytics/v3.0/sentiment",
jsontext = Text.FromBinary(Json.FromValue(Text.Start(Text.Trim(text), 5000))),
jsonbody = "{ documents: [ { language: ""en"", id: ""0"", text: " & jsontext & " } ] }",
bytesbody = Text.ToBinary(jsonbody),
headers = [#"Ocp-Apim-Subscription-Key" = apikey],
bytesresp = Web.Contents(endpoint, [Headers=headers, Content=bytesbody]),
jsonresp = Json.Document(bytesresp),
sentiment = jsonresp[documents]{0}[detectedLanguage][confidenceScore] in sentiment

 

I then go to my query with the data to analyse, Invoke Custom Function - choose the above function but the returns the results as an error:

 

Really hoping someone's got some advice because this is breaking my brain!!

 

 

 

 

  • amitchandak just wanted to give you an update, I've finally managed to get the sentiment analysis to work!!

    I created parameters for the code so it ended up looking like this:

     

    (text) => let
    apikey = APIKey,
    endpoint = Endpoint& "/text/analytics/v3.0/"&Service,
    jsontext = Text.FromBinary(Json.FromValue(Text.Start(Text.Trim(text), 5000))),
    jsonbody = "{ documents: [ { language: ""en"", id: ""0"", text: " & jsontext & " } ] }",
    bytesbody = Text.ToBinary(jsonbody),
    headers = [#"Ocp-Apim-Subscription-Key" = apikey],
    bytesresp = Web.Contents(endpoint, [Headers=headers, Content=bytesbody]),
    jsonresp = Json.Document(bytesresp)
    in jsonresp

     

    I also had a parameter for language replacing the ""en"" but this didn't work so reverted but to the orginal

    I'm so happy it worked in the end!!

6 Replies

  • PaulaMcC , Click on the error, You will get some text. Share that here. See if that can give you some hint where is the error.

     

     

  • PaulaMcC's avatar
    PaulaMcC
    Frequent Visitor

    amitchandak this is the erro:

    An error occurred in the ‘’ query. Expression.Error: The field 'detectedLanguage' of the record wasn't found.
    Details:
    id=0
    sentiment=positive
    confidenceScores=
    positive=1
    neutral=0
    negative=0
    sentences=[List]
    warnings=[List]

    • amitchandak's avatar
      amitchandak
      Super User

      PaulaMcC , Is showing an error on detectedLanguage, Have you removed any columns ??

      What are the steps you have performed

      • PaulaMcC's avatar
        PaulaMcC
        Frequent Visitor

        amitchandak I have removed columns prior to Invoked Custom Function. I've also got the text analysis custom function in the same query, I'll try it in a fresh query no & see if it makes a difference

  • PaulaMcC's avatar
    PaulaMcC
    Frequent Visitor

    amitchandak So I've tried stripping back the query so there are less steps involved but it's still returning an error.

  • PaulaMcC's avatar
    PaulaMcC
    Frequent Visitor

    amitchandak just wanted to give you an update, I've finally managed to get the sentiment analysis to work!!

    I created parameters for the code so it ended up looking like this:

     

    (text) => let
    apikey = APIKey,
    endpoint = Endpoint& "/text/analytics/v3.0/"&Service,
    jsontext = Text.FromBinary(Json.FromValue(Text.Start(Text.Trim(text), 5000))),
    jsonbody = "{ documents: [ { language: ""en"", id: ""0"", text: " & jsontext & " } ] }",
    bytesbody = Text.ToBinary(jsonbody),
    headers = [#"Ocp-Apim-Subscription-Key" = apikey],
    bytesresp = Web.Contents(endpoint, [Headers=headers, Content=bytesbody]),
    jsonresp = Json.Document(bytesresp)
    in jsonresp

     

    I also had a parameter for language replacing the ""en"" but this didn't work so reverted but to the orginal

    I'm so happy it worked in the end!!