To provide solutions for every feedback analysis use case, especially across multiple languages, Zurvey.io offers a solution via its dataset translation capabilities.
b) Translate Input Data Automatically
c) Translate Analyzed Data Automatically
d) Translate Analyzed Data Automatically With HTML Formatting
e) Combining Translation Functions
a) About Dataset Translation
This feature set aims to streamline analysis and reporting by providing two primary translation functions. You can either translate the input data for analysis or opt for unified reporting by automatically translating analyzed verbatims. Each function offers unique advantages, and they can even be combined to cater to specialized use cases. It's important to note that these translation features leverage Google's Cloud Translation service, with considerations for character quotas. The system proactively notifies group administrators at specified thresholds of quota usage (70%, 90%, 99%, 100%). For additional quota requirements, you can reach out to your account manager. Furthermore, access to these features requires group-level activation, which is available upon request. Our Customer Success Managers are on hand to assist in setting up translation quotas for your group.
Let's delve deeper into these capabilities.
b) Translate Input Data Automatically
Upon entering the Translation Settings section, you’ll encounter the "Translate input data automatically" function as a primary option (1). This feature instantly translates all incoming or uploaded data to the language specified in the Analysis Settings as the Analysis Language (2). It serves as a valuable tool for cases where analyzable feedback is generated in multiple languages, requiring unified analysis, or when Custom Labels are defined exclusively in one language. Moreover, it’s beneficial for feedback generated in languages not currently supported by Zurvey.io's native language analysis. See the available languages here.
Text units are only translated if their language differs from the analysis language, hence translation quota will only be counted for text units of different languages. The original input data and its language can be found on the dashboard and in the exports.
The drawback of this method is that some information may be lost in translation, compared to analyzing the original language.
Before activating this function, please consider the following limitations:
- Input data translation is currently only available for NMI integration datasets and File upload datasets.
- Translation may increase data processing time.
- If the analysis language is set to Mixed language, the input data translation feature will be disabled.
- Ensure that you set up the correct Analysis Language, as changes cannot be made later.
- Once input translation is activated, it cannot be deactivated. Additionally, datasets created without this function being active cannot be modified later. Changes regarding this setting cannot be made during reanalysis either.
Please also consider the following regarding Translation quota management:
- Data processing will halt if translation quota is exhausted. Processing will resume once sufficient quota becomes available. If you do not wish for these analyses to restart, it is advised to archive these datasets before adding additional quota.
- If archived datasets are restored, the input translation feature will automatically resume translation processes as well if sufficient quota is available.
c) Translate Analyzed Data Automatically
The second translation option is to automatically translate analyzed data. When activated, this feature translates the analyzed data into English, while keeping the original language for the analysis itself. This means that only the results are translated. The dashboard then displays both the original text and its English translation side by side. This translation applies not only to individual text entries but also to labels. Additionally, exports like Semantic and Result Exports include an extra column for these translations.
It's worth noting that this feature can be used alongside Input Data Translation. This setup suits cases where you want to analyze feedback in one language but reporting and visualization should happen in another. For example, if feedback comes in multiple languages and you've created labels in German, you can translate the data for analysis, but still visualize it in English on the dashboard.
d) Translate Analyzed Data Automatically With HTML Formatting
Once this feature is enabled, analyzed data is automatically translated to English while preserving HTML formatting for verbatim visualization, both in the feed and in drill downs. This ensures that phrases with sentiment value or recognized topics are highlighted in the translated text, just as they were in the original analysis. This improves the clarity and transparency of text analysis results, making them easier to understand even for those unfamiliar with the original language.
It's important to note that when Zurvey.io sends verbatims to the translation service with HTML formatting, it may consume approximately three times more translation quota compared to instances without formatting.
e) Combining Translation Functions
As mentioned earlier, you can combine translation functions. Let's look at an example: suppose the feedback is in French, and both input translation and analyzed data translation are activated, with the analysis set to German.
In this scenario, the system first translates the original text to German, and then translates the analyzed German text to English. As a result, the data analysis occurs in German, but the dashboard automatically displays English translations alongside the original texts. This translation applies to both text units and labels. Additionally, Semantic and Result Exports include extra columns for both the original text and these translations.
Read more about this feature on our blog: