Error Handling Azure Content Moderation API

Question

Error Handling Azure Content Moderation API

Reinhardt Erasmus 5

I'm trying to get create a AuditLog for my moderated content and would like to add the Exception Message sent by the Azure Content Moderation API to the log. Can anyone assist me with regards to saving this to a variable?

This is what I got from some forum and googling but its not exactly what I am looking for.

string exceptionMessage = screen?.Classification.ReviewRecommended ? "Azure review recommended" : "Custom warning message"; // Adjust this as needed based on Azure's response.

Reinhardt Erasmus 5

To give you context to how it will work:

if (isProfanity)
                {
                    
                    throw new RemoteException("The username contains profanity. Please pick a different one.",
                        "The username contains profanity and has been denied.",
                        RemoteExceptionType.ProfileUsernameProfanityException);
                }
                else if (warning)
                {
                    var log = new ProfanityAuditLog
                    {
                        Username = text,
                        DetectedLanguage = detectedLanguage.DetectedLanguageProperty,
                        IsProfanity = isProfanity,
                        ThresholdCategory1 = _options.ThresholdCategory1,
                        ThresholdCategory2 = _options.ThresholdCategory2,
                        ThresholdCategory3 = _options.ThresholdCategory3,
                        //ExceptionMessage = exception
                    };
                    _profanityAuditRepo.AddLogAsync(log);
                    throw new RemoteException("The username might contain profanity.",
                        "The username might contain profanity and will have to be validated by support.",
                        RemoteExceptionType.ProfileUsernameProfanityWarningException);
                }

If the word is profanity the remote exception should be thrown and the word should be considered profanity, if the word falls between the ratio of 0.5 & 0.7 the word should be categorized as a warning and be added to the AuditLog for user validation and a emote exception should be thrown to tell the user. The issue is the Exception Message I am referring to in the log  needs to be the exception I receive from Azure, this is in case the issue is related to Azure and not the word e.g. API service is down.

Reinhardt Erasmus 5 Reputation points

2023-08-11T09:14:20.4166667+00:00

@romungi-MSFT Thanks for the feedback, I'll try implementing it like you suggested, think your suggestion to just return a request failed if no JSON is received should work, I will then just implement a custom response that states it was a request failure.

Just needed to confirm if there wasn't possibly a call or method I could use to get a exception directly from the API itself.
romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-13T15:01:08.88+00:00

Sure thanks!!

1 answer

Your answer

Reinhardt Erasmus 5 Reputation points

2023-08-10T14:40:09.4966667+00:00

To give you context to how it will work: if (isProfanity) { throw new RemoteException("The username contains profanity. Please pick a different one.", "The username contains profanity and has been denied.", RemoteExceptionType.ProfileUsernameProfanityException); } else if (warning) { var log = new ProfanityAuditLog { Username = text, DetectedLanguage = detectedLanguage.DetectedLanguageProperty, IsProfanity = isProfanity, ThresholdCategory1 = _options.ThresholdCategory1, ThresholdCategory2 = _options.ThresholdCategory2, ThresholdCategory3 = _options.ThresholdCategory3, //ExceptionMessage = exception }; _profanityAuditRepo.AddLogAsync(log); throw new RemoteException("The username might contain profanity.", "The username might contain profanity and will have to be validated by support.", RemoteExceptionType.ProfileUsernameProfanityWarningException); } If the word is profanity the remote exception should be thrown and the word should be considered profanity, if the word falls between the ratio of 0.5 & 0.7 the word should be categorized as a warning and be added to the AuditLog for user validation and a emote exception should be thrown to tell the user. The issue is the Exception Message I am referring to in the log needs to be the exception I receive from Azure, this is in case the issue is related to Azure and not the word e.g. API service is down.
Reinhardt Erasmus 5 Reputation points

2023-08-11T09:14:20.4166667+00:00

@romungi-MSFT Thanks for the feedback, I'll try implementing it like you suggested, think your suggestion to just return a request failed if no JSON is received should work, I will then just implement a custom response that states it was a request failure.

Just needed to confirm if there wasn't possibly a call or method I could use to get a exception directly from the API itself.
romungi-MSFT 48,906 Reputation points Microsoft Employee Moderator

2023-08-13T15:01:08.88+00:00

Sure thanks!!

Answer 1

I think in the case where the service could be down there will be an error message from the moderation API and there would be no JSON to check the profanity score, so I think in this case you can add the response received as request failed to your audit log.

For the scenario where the exception needs be differentiated, I do not see a configuration setting that could help to differentiate them based on the profanity score. I think client-side processing needs to be done to segregate the same.

If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

Share via

Error Handling Azure Content Moderation API

1 answer

Your answer