Introduction
Usage
Tutorials
Introduction


                                                                                                        【中 文】


Sentence Generating Function using Neural Network

MadrasDataListIcon A language is composed of words and sequences of words. When we listen to a talk or read an article, we get information not only from individual words but also from the sequence of these words. If we want to use neural networks to process a language, a network with memory which can process temporal sequences, would be a better choice. Therefore, we use DFA(Discrete Finite Automata) networks to process and model a language. By taking advantage of DFA's ability of predicting the next word, we can input the user's query sentence to the trained DFA network and do a similarity comparison for semantic search with the query sentence plus the next word in the sentence.

We also include the results by NMF(Non-negative Matrix Factorization) and SOM(Self-Organizaing Map).

We use 16 works of Mark Twain as our training corpus.

  1. The Innocents Abroad.
  2. Roughing it.
  3. The Adventures of TOM Sawyer.
  4. A Tramp Abroad.
  5. The Prince and the Pauper.
  6. Life on the Mississippi.
  7. The Adventures of Huckleberry Finn.
  8. A Connecticut Yankee in King Arthur's Court.
  9. A Horse's Tale
  10. Christian Science
  11. Extract From Captain Stormfield's Visit to Heaven
  12. Is Shakespeare Dead From My Autobiography
  13. The American Claimant
  14. The Mysterious Stranger
  15. The Tragedy of Pudd'n'head Wilson
  16. Tom Sawyer Abroad

We use these three networks to obtain semantic codes, and use these codes to accomplish semantic analysis.

MadrasDataListIcon Download Semantic Indexing on Mark Twain's works Demo

MadrasDataListIcon Program Usage 

MadrasDataListIcon Tutorials

MadrasDataListIcon Any questions, please contact with cyliou@csie.ntu.edu.tw

MadrasLine
All Rights Reserved by Cheng-Yuan, Liou,
Department of Computer Science and Information Engineering,
National Taiwan University.