KantanAnalytics™ - MT Quality Estimation

KantanAnalytics™ generates quality estimation scores for automated translations generated by KantanMT engines. These help Project Managers predict the cost and post-editing effort for Machine Translation projects.

KantanAnalytics, which has been co-developed with the CNGL Centre for Global Intelligent Content (Dublin City University, Ireland) assigns a quality estimation score - expressed as a percentage - for each automated translation generated by a KantanMT engine. Segments with higher KantanAnalytic scores have proven to require considerably less post-editing, reducing post-editing time and cost!

 

KantanAnalytic Reports are designed for Project Managers - helping them to predict the post-editing effort required to complete a project and to determine how long it will take!

KantanAnalytics Report

KantanAnalytics creates a detailed project management report of all segments within a KantanMT project. This includes segment-by-segment quality estimation scores in addition to other useful project statistics such as word, character, placeholder and tag counts. Project Managers can use this information to :-

  • Develop a Tiered Cost Model - Tied to the KantanAnalytics quality estimation scores, higher scoring segments can be charged at a lower rate. This creates a tiered business model which more accurately predicts the actual cost of post-editing for a machine translation project.
  • Predict Project Schedule - As higher scoring segments are proven to take less time to post-edit, a more accurate and precise project schedule can be determined using the KantanAnalytics statistics.
  • KantanAnalytic Reports - These reports provide detailed analysis of the performance of KantanMT engines - including segment level quality scores, word, character, tag and placeholders counts. Think 'Fuzzy Match' report for MT and you've got the general idea!
  • Open Standards - KantanAnalytics generates an XLIFF document containing both source and target translations for each processed file. Use these in CAT environments to start working with the quality estimation scores immediately!
  • Monitor Quality Performance - The better the KantanAnalytics scores - the better the quality performance of a KantanMT engine. An engine with a high distribution of low-scores should be retrained or replaced with a better more domain specific engine. KantanAnalytics can help Project Managers make the right decision.

Contact Us