The environment “Mnemosyne” by Neurolingo is used in the TT Hellenic Postbank data cleansing project

The TT Hellenic Postbank has chosen the environment “Mnemosyne” developed by Neurolingo for using it in a new project aiming at the data cleansing of its products and accounts.

The environment "Mnemosyne" is a complete language data processing environment that incorporates all the technologies and dictionaries developed by the company. The task of data cleansing requires the matching of records that may be related to the same entity and therefore should be combined, completed, corrected or eliminated. The analyzers of the Mnemosyne system, based on methods of natural language processing (NLP), allow the matching of linguistic data, and more specifically the matching of Named Entities, such as person names (surname, first name, father name, mother name), dates of birth, addresses, etc.

To meet the specific needs of this project, Neurolingo will operate customization and optimization of the “Mnemosyne” in order to generate the best possible use of the system and obtain the best possible results. Some of the actions that will allow overcoming linguistic problems in the framework of this project are the following:

  • a) normalization of noun forms based on both morphological/declination information (e.g. forms in genitive such as in the field ) and grammatical rules; in particular, the identification of each name (of person or place) will take into account its several graphical versions (transcription of Greek names with Latin characters, transcription of foreign names with Greek characters, abbreviations, etc.) as well as relations between synonymous expressions (for instance, children that have not yet been baptised – hence, their Christian name cannot be used in official records – are now recorded under more than 60 different expressions in the TT database).
  • b) development of special dictionaries for Greek names (male and female, first/Christian names), names of roads, cities, counties, regions, countries, etc.