A neuron model for documents containing multilingual Indian texts

A two input neuron model for multi-lingual web pages for education Indian subcontinent is an area of multilingual and multi-cultural diversity. Introducing web-enabled education is a challenge as one needs content-based approaches for web pages with english and regional language texts. Translational tools many times get hampered due to time and context-based interpretations. So a generic tool to identify the content of web pages is needed and this will form the basis for later elaborations on the text. The aim of the paper is to present a 2-input neural model which captures salient features of a subject to identify the content in a regular web page containing both English and Telugu-one of the regional languages of India- and test it with computer generated, printed and handwritten formats. Typical words having common content are chosen and neural network is used to arrive at a normalised output. The example of a physics text discussing magnetism is used as an illustration.