Turning Queries into Knowledge - GALA Global ·  · 2017-07-13Actors Sources: Google image search:...

Post on 06-May-2018

216 views 2 download

Transcript of Turning Queries into Knowledge - GALA Global ·  · 2017-07-13Actors Sources: Google image search:...

Turning Queries into Knowledge Or rather:

A More "Holistic" Approach to Vendor Quality?

Anita Wilson, eurocom Translation Services, Vienna

Klaus Fleischmann, eurocom & Kaleidoscope , Vienna

• Multi-language vendor

• Vienna, Austria

• 16 project managers

• Translation exclusively outsourced

• Technology-oriented

• Language Technology

• SDL reseller & tech partner

• Own software products for “fringe processes” – In country-review

– Query management

– Terminology workflow

– Small problem-solvers

Actors

Sources:

Google image search:

blog.phpforms.net

www.xavierleadershipcenter.com

www.richardsolo.com

001yourtranslationservice.com

Why Quality – Our Strategy

• Quality will survive for human translation

– Techdoc is decreasing

– Clear best practices, tough USP

• Measurable quality CAN BE a USP

• For MLVs, vendor quality is key

• Query management discussion 2011

eurocom´s Approach So Far

• EN 15038-certified

• QA according to SAE J2450

• Dedicated Vendor Management

• Own VM-Tool

• Vendor selection & classification

• Weekly meetings

QA Assessment of Translations

VM Tool

Weekly Production Meetings

• PMs share experience with Vendors

• VM takes action points and updates VM Tool

Status Quo Quality Workflow

• EN 15038

• QA checks

• Query management

• In-country review

• QA Assessment carried out in certain cases only

Downside of Current Approach

• Real investigation only on a per-project basis

• No consistent and objective data:

– No historic data

– Data not in relation to amount of jobs etc.

– No objective input from multiple sources

• Many subjective and preferential opinions

Visions

• On-going, historical evaluation of vendors

• Use existing data sources and make them

– Objective

– Comparable

• More “holistic” view on vendor quality

0

2

4

6

1 3 5 7 9 11

Possible Sources of Data

• Queries

• In-country Reviews

• Terminology requests

• QA Checks

• Hard facts

• Soft facts

• Informal input

Data Source: Translator Queries

• How many and what queries?

– Terminologically relevant

– Actual source text errors

– Silly

• Statistical data fed to VM tool

• Turn queries into know-how

– For quality and for future projects

Workflow & Re-usability

• Tracked and checked for implementation

• Centrally stored

• Re-used and searched automatically

• Exported to termbase

Translator (or SLV)

Project Manager

New queryAnswer

smartQuerydata base

Client Contact

Subject Matter Experts

Delegate

Answer

Answer

Delegate

ClientTermbase

Categories

PMs define “silly” queries

Study Result From 2011

139

123

96

68

40

22

17

0 50 100 150

Ü prüfen

Defekt AT

Verständnis

Abkürzung

Eigenname

Feedback

StyleGuide

Issue in target term

Defective source text

Issue in source term

Abbreviation

Proper name

Feedback

Style Guide

236

168

99

49

14

9

0 100 200 300

Ü prüfen

Abkürzung

Verständnis

Defekt AT

Eigenname

Feedback

Issue in target term

Abbreviation

Issue in source term

Defective source text

Proper name

Feedback

-> Most queries are about terminology

Screenshots

Data Source: Term Requests

• What terms do translators suggest?

• Study of 2009

– Relevant for terminology

– Quality measurement?

Data Source: In-country Review

• globalReview

– Web-based in-Country review tool

– Reviewers categorize all changes

– Translators double-check changes AND categories

Example

Data fed to QA matrix and VM Tool

Data Sources: QA-Checks

• Verifika / Studio

– Map errors to eurocom quality matrix

• Problems

– How to exclude false positives?

– How to add “meaning” to the error reports

– Automated data transfer?

Verifika Screenshots

Data Sources: PM & PM Tool

• Hard facts from PM Tool

– Words per time

– Deadlines

– Completeness

• Soft facts from “Project Finalization Form”

– Allows PM to give a subjective impression

Data Sources: PM & PM Tool

Data Source: Project Managers

• Regular meetings

– Entirely project-related

– Emotional

– Not in relation to the job specifications or the quantity of jobs done

• Objective?

Bringing it all together (1)

• So what were the sources?

– QA Checks

– In-country Reviews

– Hard facts such as deadlines, completeness…

• Easy to calculate and feed to VM Tool

• Customer-specific weighting & benchmarks

Bringing it all together (2)

• So what were the sources?

– Queries

– Terminology requests

– Soft facts

– Informal input

• Data but no clear benchmarks yet

The Ultimate Vision

• Everything in one VM Tool

• Historical data and reports available on demand

• PM and VM (and end clients) receive an accurate picture of Vendor quality…

0

5

10

1 3 5 7 9 11

Q&A

anita@eurocom.at

klaus@eurocom/kaleidoscope.at

@ExpressYourBiz2 #querymanagement

Expressyourbiz2

Kaleidoscope GesmbH

Kaleidoscope - Express Your Biz