GlobalSight release 8.7.3

GlobalSight 8.7.3 is the new release of the GlobalSight TMS.

For instructions on upgrading to the new release, see Upgrading GlobalSight.

New Features

Post Edit distance report

The new reporting feature will generate edit distance for MT programs. The edit distance data will help the Program Teams to monitor the MT quality and effort involved in post-editing the MT. It will support MTPE discount negotiations for both vendors and clients. The MT team will roll out the process for using this report and more detailed info will be provided on the wiki shortly. Users should have related permission to access this report.

Example

Example report, headings will be: Job id, Job Name, tu id, Post Edit %, TER %, source, MT, translated Text. A summary tab will provide a rolled up language / job summary.

Known Limitations:

1. Differentiating MT from Translated text by colours is not implemented in current release.
2. The calculated Translation Edit Rate may be not accurate for Multi-Byte languages

The TER algorithm applied in GlobalSight has some known limitations in edit count and word count, which make the results not reliable as expected. Here are two simple illustrations about the limitations:

Case 1: Wrong edit counts
MTed Segment: 3rd party software
Human Edited Segment: 3rd-party software
TER calculated: 200% because it is regarded as one insertion and one substitution after human editing.
Fomula: (1+1) Edits / 2 Words

Case 2: Wrong edit and word counts
MTed Segment: 显示器
Human Edited Segment：监视器
TER calculated：100% because it is regarded as one substitution after human editing.
Fomula: 1 Edit / 1 Word

Activity Comment Attachment Upload Check

A new option “Check for Activity Comment Attachment upload on Activity Completion” was added to the workflow activity property.

To enable this option, users need the “edit workflow” permission in the Graphical Workflow Editor. Ensure that the “Check for activity comment attachment upload on activity completion” option is enabled in any relevant activity.

When an activity has this option enabled, GlobalSight will prompt users to attach a file to this activity via the upload files before they can complete the activity or the whole workflow.

Example workflow edit activity option:

Example notification Click OK to continue the workflow to the next step without files, or cancel to cancel the workflow step completion

Files are uploaded to an activity from the following location: comments > activity comments > new

Populate MT matches option added to offline kit options

A new option, “Populate MT Target Segments”, is available for users to download offline translation kits except OmegaT bundles. They will be pre-scored at 65%. By default, this option is checked for all users. If it is not checked, the MTed target segments will roll back to original segment text and will be scored as new words.

TM option “Include MT matches as separate download file” will separate out any MT matches into a different file. If not checked, the MT matches will be rolled into a single tm file along with 100%, ICE and fuzzy matches.

MT count in word counts

Users now can get MT word count information on Detailed Word Counts web page, and from the following web service APIs*:

getJobAndWorkflowInfo, fetchWorkflowRelevantInfo, fetchWorkflowRelevantInfoByJobs, getJobsByTimeRange, getDetailedWordcounts (API for Falcon)

API reference here http://www.globalsight.com/wiki/index.php/GlobalSight_Web_Services_API

- A new button Leverage MT now is available on Job Details and Activity Details page if job’s related workflow has MT configured. - The “Re-try MT” Option has been removed from Update Leverage dialog, users now use the “Leverage MT” button instead.

- For job creation, a new option Use MT Engine is added to Localization Profile settings page. By default, Use MT Engine is set to Automatic, so GlobalSight will use MT during job creation. Set Use MT Engine to Manual to skip MT in job creation. MT can be then applied manually at the users discretion by the new Leverage MT button.

Note *: Displaying MT word count information only applies to jobs created after 8.7 upgrade. Existing jobs won’t display such information.

Job details page – note the state is “ready to be dispatched”

Word counts, successful MT count

Remove workflow % complete column in job summary page

The workflow % complete column in job summary page has been removed. Users review workflow state by the different font and colours according to the state of the workflow. Note that Workflow % complete still available by API, but may be deprecated in a future release.

8.6.7 showing workflow % complete

8.7.0 % complete removed

In Context Review tool now supports HTML files

The incontext review online editor can now render html files. To use this feature, make sure that: - Rerun GlobalSight Installer has been run, checking the In Context Review Tool on Optional Packages screen, if it has not already been enabled. - Sign in as superadmin and check HTML option beneath In Context Review on Edit Company page.

HTML: need extract content attribute values in <meta>

Associated attributes of Meta tag are configurable for translation after the change.

Examples:

1. <meta content="By Name Lastname on Mon, May 02 2016" author="Name Lastname" date="Mon, May 02 2016" name="gfeAuthor"/> (Attribute "content" and "date" are configurable for translation. Attribute “name” and “author” are always not-translatable as before.)
2. <meta content="text/html; charset=UTF-8" http-equiv="content-type" /> ("content" attribute and "http-equiv" attribute both are always not-translatable as before)

Date of deployment of patch

1. On "About GlobalSight" UI, change version info like this: Version: 8.6.9 Build 2016-05-19.
2. On "Installed Patches" UI, add date as suffix in patch list as (yyyy-MM-dd)

JBoss/WildFly Upgrade

GlobalSight’s J2E application engine is upgraded from JBoss EAP 6.4.0 to WildFly 10.0.0.Final. Java should also be updated to jdk 1.8_91. Functions deprecated.

Create Job (deprecated) 
GS Editions 
Snippet Import 
SGML Rules 
RSS Reader 
Import 
Upload 
Corpus Aligner

TM2 to TM3 conversion disabled. JMS queuing disabled in favour of our own queue Jboss J2E web profile enabled.

Add create job relevant apis to restful apis set

Six create job relevant API’s are added to the current TM API’s. More info can be found on the wiki shortly.

GET/getUniqueJobName
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/jobs/getUniqueJobName

POST/uploadSourceFile
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/jobs/sourceFiles

POST/createJob
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/jobs/createJob

GET/getJobStatus
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/jobs/{jobId}/status

GET/getJobExportFilesInZip
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/jobs/{jobIds}/targetFiles

GET/getFileProfiles
http://localhost:8080/globalsight/restfulServices/1.0/companies/{companyName}/fileProfiles

After this fix, it applies token authentication. First of all, use:

http://hostName:port/globalsight/restfulServices/companies/{companyName}/login-helper

to get an “accessToken” for valid “username” and “password”

accessToken should be passed as header value in subsequent calls after login.

XLIFF alt-trans tag not populating job TM

1. When import an xliff1.2 file, all alt-trans info will populate to leverage match table.
2. In segment editor, section "Xliff Alt" to show these "alt-trans" will not display.

Globalsight installer log

Add log support in GlobalSight installer. When run GlobalSight installer, it can log INFO, WARNING, ERROR into log file "Install.log".

maintain knowledge of which are MT segments

Users can track which segments started out as MT segments all the way from job creation to tm update, and then from within the tm itself. For tmx export write a <prop type="x-mt"/> on the translated tuv.

Ability to Identify MT Matches in Reports

If the original first leverage of a segment comes from MT, the segment will be identified as an MT Match in the TM match (original first leverage) column in the following reports:

• Comments Analysis Report
• Character Count Report
• Reviewers Comments Report
• Reviewer Comments Report (Simplified)
• Post-Review QA Report
• Translations Edit Report
• Translation Verification Report
• QA Checks Report
• Implemented Comments Check Report
  

QA Report Update – Include Initial Leverage Match

A new column named TM match (original first leverage) now is available in QA Checks reports.

Base Text Filter Permissions

A new permission labelled “Base Text Filter (View)” is added under Data Sources -> Filter Configuration (View) on the Edit Permission Group page for all Permission Groups. Which has 2 sub-permissions labelled Internal Text (Edit) and Escaping (Edit).

For permission groups that have Filter Configuration (View) enabled, the new Base Text Filter (View) permission and Internal Text (Edit)/Escaping (Edit) sub-permissions will be enabled by default on upgrade (when this feature is being deployed for the first time). For permission groups that have Filter Configuration (View) disabled, the new Base Text Filter (View) permission and Internal Text (Edit)/Escaping (Edit) sub-permissions will be disabled by default on upgrade (when this feature is being deployed for the first time).

Change to (or new) Download API to allow Combined Download

It is now possible to download all files from a job (or multiple jobs) created via API in a single zip – exactly as it is possible via the GlobalSight UI.

• Add getJobExportFilesInZip() API and getWorkflowExportFilesInZip() API in original Ambassador .Return string in XML format contains information of exported job files as above.User can be able to download all files from a job (or multiple jobs) in a single zip – exactly as is possible via the GlobalSight UI.

New GetFileProfilesForL10nProfile API call

Get file profile information according to L10N Profile name and company name. General users can get L10N profile of their own company, super PM and super admin can get L10N profile of all companies.

Comprehensive Pending Job Information

If a job fails to create, the Show Details link will appear next to failed source pages on Import Error page. Users can access the error details and/or the stack trace for each failed file.

show in editor for MT

The "Show In Editor" has been removed from mt profile.

Add TM matches info in Segment Details window

On job details page and activity details page, in View translation status, popup editor, post review editor and In Context Review page, right click one segment, you will be able to see segment details. On segment details page, details of the segment is shown.

Popup Editor to Support Viewing and Replace Picture File

Previously for pictures like png, jpg etc, GS treated them as unextracted files. Now the review of picture files is supported on the job details UI. Clicking a picture file will open it in a separate window. In this window, user can preview source/target pics and can upload to replace the target pic if it is translation activity. Pictures supports are jpg, jpeg, jpe, png, bmp, gif.

GS should log the fact that it has skipped a segment where the segment length is > 1000 characters

Max Segment Length (total segment chars) is configurable from the MT Profile setting page. If the character length of a segment is bigger than the inputted one, it'll not be sent to MT with logging it in the globalsight.log as below:

2016-04-01 15:48:30,708;[pool-5-thread-1];INFO ;com.globalsight.machineTranslation.AbstractTranslator;Max chunk length (2) exceeded; the segment will be ignored. Segment info: length = 59; text = <bpt i="1" type="x-span" x="1"/>Sample Document<ept i="1"/>

make MT char limit configurable

MS translator have changed their limits on max string length. We have allowed to make the max length string to send configurable as follows:

• 1. Label "MT Confidence Score" is changed to "MT Threshold Level" as in GBS-4305.
• 2. "Show In Editor" is removed as in GBS-4276.
• 3. New option "Ignore TM Matches for MT": It only be able to be checked when MT leverage threshold is set to 100%.

       If it's checked: When creating job, it'll not leverage from TM. So all segment will be sent to MT. 

• 4. New option "Log Debug Info". It controls logging of MT debugging info in globalsight.log.
• 5. New options of MS_Translator:

       a. "Translation Mode": 
               "Send Text Between Tags" only send text between tags to MS_Translator; 
               "Send Text Including Tags" send text including tags to MS_Translator; 
       b. "Max Chunk Length (total segment chars)": Default is 1000. Max length of characters to send to MS_Translator. 

• 6. All MT'ed segments will be saved to TUV table regardless "MT Threshold Level", and are translated.

MT confidence score label incorrectly named

Relabel "MT Confidence Score" to "MT threshold Level", in the reports are "MT Threshold Level" too. Corresponding reports are "Online Jobs For IPTranslator Report " and "Detailed Word Counts Report".

Force MT

Add a new option "Save MTed Segments to TM " on the page "TM Profile" to decide whether to save un-touched MT segment to TM or not. During populating to TM, a new property "<prop type="x-mt">MS_Translator_MT</prop>" will be added in the TUV.

Known issue: If segment is translated by user "xxx_MT", then it'll be recognized as MT'ed segment, and be populated to TM with the new option is checked.

API create job file profile

Globalsight will issue a warning if invalid data is used to create a job via API. Even though this warning is issued, the job will still create as normal. However, we do have it on our roadmap for a few release’s time, to fail job creation if invalid data is entered.

Dynamic Quality Framework fields (DQF)

Dynamic Quality Framework (DQF) fields are now available as part of the existing Scorecard feature. The Scorecard feature provides a mechanism for the reviewer to give feedback about the translation. DQF allows more detail to be added. For the reviewer, there are some extra fields: Fluency, Adequacy, Comment, and segment level Field Severity in addition to Category failure in the offline Reviewer's comment report.

Example of a revirwer's comment report

Segments Category and Severity, online example

Online Fluency and Adequacy summary

The scorecard tab in the Job Details page for Project Managers

To turn on DQF, enable it in the workflows that require DQF input

Wild card search in TM search

Using this advanced TM search option, it is now possible to search with a blank source and target field, This feature is only available in the advanced search options. At least one other option must be filled in, e.g. SID, create date. The return is limited to 300 results.

You can enable this feature in the company page

Example of returns

PO files - Context and Translator notes info as reference

Within PO files, the fields Context and Translator notes are extracted and can be displayed inline. Note that these should not be translated and should be set as internal text. GlobalSight will not accept changes for these segments.

Plugin for Adobe Experience Manager

A plugin for Adobe Experience Manager (AEM) is now available upon request.

Addition of Secondary JSON filter to XML filter

For the XML filter, once a selection of text has taken plece, there may be a requirement to process the selected text further. This enhancement allows a secondary filter of JSON type to be applied to the output of the XML filter.

QA Checker - differentiate source==target types

There are two types of source==target segments, approved and untranslated.

One types is generated when the translator has approved the source==target and the other type contains untranslated segments, where the translator has not toughed the segments. The QA report will now differenciate between the two segment types.

Example of a QA report.

Microsoft Translator API change

Microsoft have announced a change in the way that the MT API’s are called in their MT service. This change needs to be done before April 2017. We have enabled both MT API’s for now.

The Client ID and Client Secret will work until April 2017. An Azure subscription key will need to be obtained to continue service after this date.

Note – Client secret and Client ID will be deprecated in a future release.

Example 8.7.3 MT screen.

Overdue date to be set as 'time to accept' + 'time to complete'

For any particular activity, time allowed to accept the activity and the time allowed to compete the activity are settable. Once these timers expire, the activity is flagged as overdue. This enhancement closes a loophole whereby it was possible to extend the completion date by delaying acceptance of the activity.

Post Edit Distance report – all / latest option

For the Post Edit distance report, some users require all the segments and their history to be in the report. Others only want the latest. This enhancement allows users to select which option they require when generating the report. Note that the summary tab will generate results only from the latest segments, irrespective of which option is chosen.

Password complexity

When this security feature is enabled in the company page, this feature enforces complex password requirements. For the first turn on, simple passwords are still allowed for 3 logins, after which users will be denied access.

Disable upload of dll and exe file types

When this security feature is enabled on the company page, this feature will deny upload of any file type that is exe or dll.

Restart Globalsight from the GUI

The system administrator will now be able to restart Globalsight from the GUI.

Perplexity score MT segments

This feature is turned on via the company page. It also requires a remote service developed by Welocalize. Background: Sometimes Machine Translated segments are good quality, sometimes less so. This can vary from segment to segment, language by language. Now, by using a remote service developed by Welocalize, segments can be individually “scored” on perplexity using language models.

For flow of MT segments, GlobalSight carries out the following operations:

• Receives file, parses to extract segments
• Leverages against Translation Memory
• Sends segments that fall below the MT threshold to the MT remote service.
• Receives back MT’ed segments
• Sends to perplexity service
• Receives back perplexity score.
• Shows results to vendor
• Report on results in the Post Edit Distance report.

Any segments that have passed perplexity (low perplexity score) will now receive a leverage boost of 5% - so any segment that is 65% is a MT segment, perplexity passed, and any segment that is 60% is MT segment, perplexity failed. It will be up to the vendor to manage their review of 65% segments.

Base Text Escaping updates

Improve options in the "Base Text Escaping filter". The new options would give users the ability to choose whether or not escaping should be applied, based on where the content resides within the source structure.

The types of content this may be applied to are:

• CDATA.
• XML or HTML node/element.
• Begin\end with defined pattern or characters (RegEx to be supported).
• A XML attribute value.

The new options would be able to specify the character that denotes an escape. Known issue: Part of html node does not work for office file and CDATA content.

Import / export configurations

Further updates made to enable import/export GlobalSight configurations.

Enabled under permission: Data sources / Allow Configuration Export-Import

List of exportable / importable settings:

• Locale pairs
• Machine translation profiles
• Filter configurations
• Workflows

This list will be expanded in future versions of GlobalSight.

Deprecate old tuv matches

Some users require that TM 100% matches need review if they are over a certain age. This feature in the TM Profile page will enable a penalty on old MT matches if the metadata indicates that the TM match is over a set number of days older than the current date.

Enhancement Requests - Full List

• maintain knowledge of which are mt segments
• Ability to Identify MT Matches in Reports
• QA Report Update – Include Initial Leverage Match
• Base Text Filter Permissions
• Change to (or new) Download API to allow Combined Download
• New GetFileProfilesForL10nProfile API call
• Comprehensive Pending Job Information
• New SmartBox UseCase
• show in editor for MT
• Add tm matches info in Segment Details window
• Popup Editor to Support Viewing and Replace Picture File
• GS should log the fact that it has skipped a segment where the segment length is > 1000 characters
• make mt char limit configurable
• Force MT
• api create job file profile
• Post Edit distance report
• Activity Comment Attachment Upload Check
• Populate MT matches option added to offline kit options
• MT count in word counts
• Remove workflow % complete column in job summary page
• In Context Review tool now supports HTML files
• HTML: need extract content attribute values in <meta>
• Date of deployment of patch
• JBoss/WildFly Upgrade
• Add create job relevant apis to restful apis set
• XLIFF alt-trans tag not populating job TM
• Globalsight installer log
• Dynamic Quality Framework fields (DQF)
• Wild card search in TM search
• PO files. Context and Translator notes info as reference
• Plugin for Adobe Experience Manager 6.2
• Addition of Secondary JSON filter to XML filter
• QA Checker:differentiate source==target types
• Microsoft Translator API change
• Overdue date to be set as 'time to accept' + 'time to complete'
• Post Edit Distance report – all / latest option
• Password complexity
• Disable upload of dll and exe file types
• Restart Globalsight from the GUI
• Perplexity score MT segments
• Base Text Escaping updates
• Import / export configurations
• Deprecate old tuv matches