Universal Machine Translation Engine for Multi-Language Processing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional machine translation technologies are limited to specific language pairs, requiring multiple translation engines and high development costs, and are prone to errors due to language identification inaccuracies, especially when dealing with multiple languages or unknown languages.

Innovation Solution

A method and device for machine translation that utilize a translation rule table with multiple translation rules across languages, allowing for multi-lingual translation without language identification, reducing development costs and computing resources, and improving translation accuracy by using a training phase to generate a multi-lingual phrase table and reordering model.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple translation engines are developed for different language pairs, then translation coverage is improved, but development costs and computing resources increase

Engineering Contradiction:

Improvetranslation coverageVSAvoidnumber of translation engines

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements a universal translation engine that can handle multiple language pairs through a single system. The engine uses language identification to detect the source language and applies appropriate translation rules dynamically, eliminating the need for separate translation engines for each language pair. This multi-functional approach maintains broad translation coverage while reducing system complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The translation system is segmented into distinct functional modules: language identification module, translation rule selection module, and translation execution module. This segmentation allows the single engine to handle different languages by selecting appropriate rules rather than requiring separate engines for each language pair.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If language identification is performed before translation, then translation accuracy is improved, but errors are introduced when dealing with unknown languages

Engineering Contradiction:

Improvelanguage identification accuracyVSAvoidtranslation reliability

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary language identification to detect the source language, but prepares multiple translation rule sets in advance for different languages. If the identified language is unknown or uncertain, the system can fall back to alternative rules or multiple candidate translations, preventing errors while maintaining the benefit of language-specific optimization.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The translation engine prepares backup translation rules and multiple candidate translations in advance for potential unknown languages. This cushioning approach ensures that even if language identification fails or encounters an unknown language, the system can still provide reliable translation results without compromising overall translation reliability.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

3Manufacturing precision

If conventional translation engines are used for each language pair, then translation quality is maintained, but computing power consumption increases

Engineering Contradiction:

Improvetranslation qualityVSAvoidcomputing power consumption

Core Design Contradiction:

Manufacturing precisionVSUse of energy by moving object

Solution Approach 1:

A single universal translation engine handles all language pairs by dynamically selecting appropriate translation rules based on language identification. This eliminates the need to run multiple separate translation engines simultaneously, reducing computing power consumption while maintaining translation quality through language-specific rule application.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The translation engine dynamically adapts its behavior based on the detected source language, selecting and applying the most appropriate translation rules in real-time. This dynamic rule selection optimizes computing resource usage by only loading and processing relevant translation rules for the current language pair rather than maintaining all rules in active memory.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10108607B2Method and device for machine translation

Publication Date: 2018.10.23 ALIBABA SINGAPORE HLDG PTE LTD

AI summary

A machine translation method includes determining source language text to be translated and obtaining a translation rule table, which has been trained in advance, that includes multiple translation rules associated with the target language text and the source language text in multiple languages; determining candidate results of the target language text; and determine the target language text to be output based on the candidate results. During the translation, a specific language of the source language text need not to be specified by a user. The implementations improve accuracy of the translation, and avoid errors introduced from the process of language identification during recognizing unknown languages. The implementations avoid developing a translation engine for an individual source language of text to be translated for a certain target language, and therefore save development costs and computing resources.