ABSTRACT Ontology is as a detailed description of things, which is able to link the properties of things as well as clearly depicted the relationship between different things. Ambiguity often occurs in machine translation, the Ontology has a good effect in eliminating ambiguity. By constructing a domain ontology dictionary, summarizing common sentence of English and Mongolian as well as using dependency parsing sentence structure to complete the English machine translation into Mongolian, it is to verify the role of the Ontology to eliminate ambiguity in terms of machine translation. A SYSTEM DESIGNThis paper makes ontology-based English-Mongolian machine translation as experimental requirements so as to eliminating ambiguity in the process of machine translation through verifying this experiment. Experimental system design principles are as follows:(1) English -Mongolian Machine Translation uses the field of ontology to express semantic center. Building the Ontology must fully understand the field. The field of vocabulary can be covered the entire field. Ontology must include all common features of the field so as to playing an important role in the translation process.(2) Construction of field dictionary is an important part of the machine translation system, which contains all relevant terms in the field. Construction of the field can reflect whether the domain ontology meets its demand by constructing dictionary.(3) The English Mongolian Machine Translation System using the template translation. By summarizing the common English sentence and comparing with the corresponding Mongolian sentence, the translation template is got.

Automatic scoring system is helpful to efficiently scoring the document operational test question of MS-Office. The existed scoring methods have done a lot of valuable works. But these methods record the operation track and lead to misjudge when modifying operations. In this paper, we propose a new method for this problem. The basic idea of this method is using Office Open XML file. First, extract the corresponding XML file from MS Office 2007 document of standard answer and MS Office 2007 document of student's answer. Next, score the student's MS Office 2007 document by comparing the similarity of two XML documents. Using Windows as our experiment platform, we did a lot of tests. The experimental results show that the error evaluation rate of this system is less than 3% compared to the manual scoring.