An Analysis on Challenges & Strategies In Developing Multilingual Software For Mobile Technologies |
This paperdescribes experiences in designing and developing the software applicationsIndic languages targeted at mobile devices at CDAC Noida. CDAC has vastexperience in multilingual computing and has also implemented several projectsusing browser based architecture. Now as we go further there is increasingtrend towards browsing web through mobile devices in the remote places andwhile on the go, the Indic languages need to be tested out on various browsersfor the performance. As a start a bilingual dictionary developed for differenttarget platforms was ported on different mobile devices and experience was goodas it is not easy for a developer to take off and start developing immediately.The environment requires lots of settings to successfully port and runapplication which have some language specific module. A lot of work needs to becarried out to make successful migration of useful applications and Indiclanguages websites accessible for masses on mobile devices. We evaluatedifferent architectures to recognize multilingual speech for real-time mobileapplications. In particular, we show that combining the results of severalrecognizers greatly outperforms other solutions such as training a single largemultilingual system or using an explicit language identification system toselect the appropriate recognizer. Experiments are conducted on a trilingualEnglish-French-Mandarin mobile speech task. The data set includes Googlesearches, Maps queries, as well as more general inputs such as email and shortmessage dictation. Without pre-specifying the input language, the combinedsystem achieves comparable accuracy to that of the monolingual systems when theinput language is known. The combined system is also roughly 5% absolute betterthan an explicit language identification approach, and 10% better than a singlelarge multilingual system. The coexistenceof Western and Eastern languages and cultures poses a true challenge for globalmobility and communication in business and personal life. In this paper, wewill describe mobile software applications that are built on top ofmultilingual and cross lingual technologies for overcoming language barriers,e.g., between English and Chinese, or German and Chinese. The mobileapplications are interactive language guides, culture guides and countryguides. They support automatic translation and dialogues between people andlink the language part with information needed in a specific situation.Furthermore, users can integrate pictures into their phrases for extending thedescriptive capabilities of the dialogue component. Semantic search facilitatesaccess to words, phrases and information content.