PHP Classes

File: Docs/doc3.txt

Recommend this page to a friend!
  Classes of Guillermina Gonjon   PHP Sentence Tokenizer   Docs/doc3.txt   Download  
File: Docs/doc3.txt
Role: Documentation
Content type: text/plain
Description: Documentation
Class: PHP Sentence Tokenizer
Parse sentences and extract their word features
Author: By
Last change:
Date: 2 years ago
Size: 2,351 bytes
 

Contents

Class file image Download
Remembering / Forgeting " Forgeting is important to keep things relevant and fresh. It allows to speed out the process and save memory Memory = Word Bank Different levels of memory -> * Fresh Memory ( quick access memory ) * Permanent Memory ( Hard to forget ) * Temporary Memory ( Cash memory, it needs more re-inforcement to become permanent ) * Bare Memory ( Incomplete, briefly touched; it may dissapear at any moment if not confirmed ) * Forgotten: dumped( Is there, but it won't retrive unless a significan part is re-activated or re-learned ); Frequency: Level of frequency depends on the number of times a word has been detected within a period of time. The greatest the frequency, the easier to retrive the word and higher chances to remain on permanent memory. Every word needs a word id and frequency. The application needs a time listener to determine the frequency. Freshness: The time past since the word was last detected. Every word is kept at the top of "Fresh Memory" everytime is detected. When a new one is detected, the word past to the second placement in priority or freshness. * Divide total list by 4 ( 4 stages of memory) result is the number of placements within each memory placements = Num_items / 4 FRESHNESS = Lenght: Shorter words are easily retained in the Permanent memory than long words. Long words need more exposition in order to be remembered. Defined by the number of characters. (int) L = NUM_OF_CHARACTERS * (-1) // To keep the number low and favor less over more. Intensity: Words matching with objectives ( Grade System ) are more relevant to purpose, and therefore more desirables. The have presedence over non matching words. * Relevance shows intensity: Intensity could be either positive or negative. In this case, shallow takes the back seat. * Hight intensity words will be remembered either positive or negative. This will change the personality of the Bot. * The more possitive feedback, the nicer Bot we will have, the more negative, the most evil the Bot. For the purpose of WORD_RECORDING_VALUE, Relevance turns into a positive number; WORD_RECORDING_VALUE = FREQUENCY + FRESHNESS + LENGHT + INTENSITY