Create a Language I/O Translation Word Count Report in OSC
This article walks you through the steps of generating a Language I/O translation usage report inside of the Oracle Service Cloud (OSC) agent desktop. If you bought a package of words from Language I/O, this report shows you how many words in that package you have used so far. You can group the translations by Incident, see the OSC Threads in that Incident that were translated and also see the actual source and target content of each translation. This latter piece is important because OSC is the system of record for all content that is passed to Language I/O for translation. For security reasons, once Language I/O has translated OSC Chat or Incident content and passed it back into OSC, Language I/O can not keep a record of (or store in our database) either the source content or the translated content.
Note: Full Chat reporting is only available starting with Language I/O Chat v2.0+. (In previous versions, LNThreads are only persisted if your add-in is configured not to combine incidents after chat.) If you are on an older version of Chat, please contact your Language I/O representative to update your add-in.
Language I/O uses an OSC custom object called LNThread to store content that is passed to Language I/O for translation and the translated version of that content. This object also stores important meta data including the word count of each translation request.
You can import the report definition attached at the end of this article, or create a report from scratch following the steps below:
- Navigate to Analytics => New Report => Grid Report
- Expand All Tables => Select the LNThread table.
The fields available in LNThread are defined as follows:
Field Name | Definition |
ID | An OSC-assigned unique ID for the LNThread object |
Incident ID | The OSC unique ID for the parent Incident where this LNThread was created |
Date Created | The date and time that this LNThread object was created |
content type | A support channel identifier: 1=ticket, 2=article, 3=chat, 4=custom object, 5=social, 6=bot |
thread Type | Specific to chat (3) and ticket (1) content type records identifying: 1=agent untranslated response, 2=agent translated response, 3=agent retranslated response, 4=customer thread translation, 5=customer thread retranslation, 6=customer thread partial translation |
word count | The number of words that was translated (if any) for this LNThread record. LNThreads with the thread Type field set to 1 or "agent untranslated response" won't have any words translated. In this case the LNThread object is simply used to store the agent's pre-translated response so there is a record of it. |
translation ID | This is a Language I/O id that uniquely identifies each translation request in our system. |
Text | The text of the LNThread. See the thread Type field to understand whose text this is. |
statusMessage | The status of the translation. To ensure you are not counting unsuccessful translations add a "SUCCESS" filter on this field. |
sourceLNProjectID | This is a Language I/O ID for a project record, which associates a translation and a retranslation for the same content. **Not always populated. |
sourceLNThreadID | If this is a translation thread, this contains the ID of the LNThread that holds the source - or untranslated - content. **Not always populated. |
source ThreadID | If this LNThread holds the translation of a System Thread (i.e. a customer email), then the source content will live in that System Thread, not in an LNThread. This field holds the Thread ID for the System Thread in question. **Not always populated. |
- We recommend that you include the following fields in your report by dragging them from the LNThread table onto the Data Area: Incident ID, translation ID, Date Created, wordCount, threadType, contentType, statusMessage and Text as shown in the example below.
- It is helpful to add filters to better sort your results. Drag fields that you want to be able to sort on at runtime to the Docked Filters area. For a word usage report, you'll definitely want to add Date Created to Docked Filters and per below, select "Make this filter selectable at runtime." and set the Operator to "between" so you can select a date range.
Lastly, to get rid of 0 word count values, you can just create an invisible, permanent filter for the agent-untranslated-response thread type. These will always have a zero word count as this is content that hasn't yet been translated. To create this filter, drag the threadType field to the Docked Filters area in Design View and create a filter that removes all LNThread records from your report where the threadType value is set to "1" as shown in the example below.
- Hit OK on the new filter and save your report to a folder in Reports Explorer. When you run your report, be sure to properly set your date range. If the package of words you are using was purchased from Language I/O on January 6, 2020, you'll set that as the start date and set tomorrow's date as the end date to be sure you get all translations to date.
- Now you can select the Export button from the navigation ribbon above your report and export the contents into an Excel spreadsheet where you can sum up your wordCount field and perform other analytics on your translation data as needed.
Comments
0 comments
Please sign in to leave a comment.