Research On Transliteration Method Of Spanish Place Names Based On Knowledge Graph

Geographical names,the exclusive names of geographical entities,are important elements of geographic information,representing a certain time and space scale,with historical,cultural and other meanings.With the introduction of "One Belt,One Road" and its further development in recent years,China is becoming more and more closely connected with other countries in the world,playing an increasingly important role in national relations,and the importance of geographical names has emerged,and the demand for geographical names data outside China has expanded.The continuous acceleration of the modernization process of the country,the deepening cooperation among various countries in various aspects such as trade and culture,the more advanced production level and abundant talent reserve have accelerated the process of global basic geospatial information construction,and the country launched the global geographic information resources construction project in 2015.In this paper,we evaluate the current methods of translating geographical names according to the technical regulations of the Technical Regulations for the Production of Geographical Names Data for the Construction and Maintenance and Update of Global Geographic Information Resources and the requirements of the project "Production of Geographical Names Data for the Construction and Maintenance and Update of Global Geographic Information Resources".The conventional method of translating geographical names is translation by experts in this field,which is time-consuming,laborious and subjective,and not suitable for translation of large quantities of foreign geographical names.Toponym translation is also different from traditional machine translation,which has exclusive translation rules and has to strictly rely on translation rules for translation.Knowledge graph technology uses graph model to represent knowledge.Translation rules are numerous and complex in structure,which can be expressed uniformly by using knowledge graph technology,and the structure of the graph model can perform knowledge inference on the constructed rules and check the consistency.In this paper,we combine knowledge mapping technology and toponym translation rules to carry out large-scale automated machine translation of toponymic monikers,improve translation efficiency while ensuring translation quality,and complete the project of "toponymic data production" with high standard.In this paper,the knowledge related to the translation of geographical names in the application manual of the national standard GB/T 17693 "Guidelines for Transliteration of Chinese Characters in Foreign Language Geographical Names",the rules for translation of Spanish geographical names,the rules for syllable optimization,the phonetic table,the rules for syllable syncopation,the geographical names in publicly published gazetteers,the names in the World Dictionary of Translation of Personal Names and various authoritative dictionaries are organized and defined as a priori knowledge.Considering the characteristics of many kinds of a priori knowledge,the existence of multiple structural forms and the existence of semantic relations,this paper constructs a knowledge map of a priori knowledge of Spanish toponymic proper names by means of knowledge classification,knowledge extraction,knowledge storage and consistency check.In this paper,the names to be phonetically translated can be divided into instance names and unregistered names according to the existence of existing Chinese translation results,while unregistered names refer to the first occurrence of the need to obtain Chinese translation results through the phonetic process.The knowledge graph-based method for instance onomastic transliteration is to conduct a search in the knowledge graph,construct inference rules for instance onomastic names,and perform rule inference on the retrieved results to determine the selected transliteration results.The phonetic translation method for unregistered names based on the knowledge graph mainly includes three parts: syllable optimization,syllable slicing,and Chinese character translation.The attributes of the unregistered names are modeled,and the modeled attributes are combined with the knowledge graph for knowledge retrieval and knowledge inference to optimize the syllables that meet the rules.This paper conducts experiments on nearly 20,000 data from Argentina,compares the experimental results with Baidu translation and only using two-way maximum matching phonetic translation,and compares the experimental results with a variety of different evaluation indexes,and the results show that the translation results of this paper’s method are more scientific and accurate,effectively improving the efficiency of place name translation and realizing the mass production of place name moniker phonetic translation.