ALIMbot ActiveX Control Specifications

VERSION: 1.0
UPDATE: 06th December 2003

Important Note

ALIMbot consists of two important parts.

1. The ALIMbot Engine which is in executable form, and 
2. The Data files of ALIMbot in ASCII following a particular format similar to that of an INI or INF file.
3. Important Notes

ALIMbot Engine

TYPE: ActiveX Control 
PROGRAMMING LANGUAGE: Visual Basic 6.0

The latest ALIMbot Engine can be DOWNLOADED from here.

The ALIMbot Engine comes as an Win32 executable. It incorporates certain features which are needed for CLEAR understanding of the user input and EFFECTIVE communication of the bot output.

Pre-processing of user input

• [IMPLEMENTED] Removal of Multiple white spaces between words in the user input
  Eg. "WHAT IS    YOUR             NAME?" would be converted into "WHAT IS YOUR NAME?"
• [IMPLEMENTED] Removal of Punctuation marks/Special characters in ASCII format in between words in the user input
  Eg. "HI, HOW ARE YOU?" would be converted into "HI HOW ARE YOU"
• [IMPLEMENTED] Removal of certain words which would be useless for processing of input
  Eg. AND, HEY etc
• [IMPLEMENTED] Replacement of certain words in user input with certain appropriate words.
  Eg. "whats up" =>"what is up" |  "u" => "you" | "howr u" => "how are you" |
• [IMPLEMENTED] Replacement of certain Question with a more appropriate question which is there in the database already.
  Eg. "who are you" => "what is your name"

Processing of user input for response

To obtain a suitable response to the user input depending on the context of previous conversation and also independent analysis.

• [IMPLEMENTED] Pattern matching of the user input with the keytags.
• [IMPLEMENTED] Allotment of a Relevance Level to each matched pattern. The hierarchy for the pattern-matching is dependent on the keytags being operated upon. The hierarchy is as follows,
  
  Keyword Hierarchy
  
  keywordonly         ^    highest relevance level
  keywordvariable   / | \
  keywordlistorder    |
  keywordlist         |
  keywordstart        | --\ lowest relevance level
  keyword             | --/ (shared by "keywordstart"
                             and "keyword" tags)
  
  
• [IMPLEMENTED] Choice of the best possible answer is done on the basis of the highest relevance level with few exceptions.
  For Eg. If more than one answers are available and one or more of them have a relevance 5001 means that answer is from [response:failure] in keywordvariable tag. i.e. the answer having highest relevance is not correct but it is the one which would be answered. Hence if more then one answers are  available along with a 5001 level answer then this relevance level is replaced by that of previous matching pattern and further comparision is done.
• [IMPLEMENTED] Search technique for keyword matching in a flat text file, termed as a AMC file, with lines of text on a particular topic. The format for this file is based again on INI type file. It's reference will exist in the data.dat file.

Post-processing of user input

• [IMPLEMENTED] Replacement of %variables with there corresponding text.

  %a	Build date (set in [general] )
  %b	Bot master  (set in [general] )
  %time	Current system Time
  %date	Current system Date.
  %u	The username
  %n	The chatbot name i.e. ALIM
  %v	The value returned from [keywordvariable] sections
  %c	Same as %v but is replaced only when the answer is a command to be executed by ALIMbot.
  %$	Inverted user input. [i->you and you->i]

   

• Understanding commands given in natural language and executing them
  Commands can be given to ALIMbot through the .dat file. If a particular pattern matches the user input then the command will trigger. This kind of structure is useful (at least till now, I think so) because it retains the flexibility and the simplicity of the .dat file and on the other hand allows different actions to be taken in the post processing section in the ALIMbot Engine.
  
  The format of the command is
  CMD::CommandName::CommandParameterArea::AnswerQuote
  
  CMD - identifies the string as a command to be processed by ALIMbot

  Variables:
  CommandParameterArea - Data passed for internal use
  AnswerQuote - is the answer to be returned to the user. If you want %v to be printed in the AnswerQuote then you have to use %c instead of %v
  CommandName - Currently ALIMbot understands the following commands 
  
  

  Command Name	What it Does?
  RUNAPP	Used for starting application like wordpad, notepad, paint brush etc when commanded. The applications to be started can be specified in the data.dat file as follows. ++++++++++++++++++++++++++++++++++++++++++++++++++ #------------------------------------------- #running applications using RUNAPP Command #------------------------------------------- [keywordvariable:run \*,run the \*,execute \*,execute the \*,open \*,open the \*,start \*,start the \*,adjust \*,adjust the \*] success=response:runok failure=response:runnotok data=data:apppath [data:apppath] wordpad=wordpad.exe text file=wordpad.exe notepad=notepad.exe internet explorer=about:blank browser=about:blank calc=calc.exe calculator=calc.exe paint=pbrush.exe explorer=explorer.exe [response:runok] CMD::RUNAPP::%v::Attempting to run %c CMD::RUNAPP::%v::Executing the command to start %c CMD::RUNAPP::%v::Hey %u! I think I started the %c CMD::RUNAPP::%v::You want me to start the %c. See %u, here it is. CMD::RUNAPP::%v::Open %c! Let me do that for you [response:runnotok] I tried but couldn't. Sorry. I don't know where it is. Sorry.. try to search for it manually. ++++++++++++++++++++++++++++++++++++++++++++++++++ %v - you know about it from the previous section - here %v would contain the name of apppath (i.e. wordpad.exe or notepad.exe etc)
  SAVE	Used for saving user entered information in the USERNAME.MEM file created by ALIMbot. ALIMbot can be commanded to save the input thru the conversation data.dat file as follows. ++++++++++++++++++++++++++++++++++++++++++++++++++ #--------------------------------------------------- #Saving User Entered Information using SAVE Command #--------------------------------------------------- [keywordlistorder:my*is*] CMD::SAVE::(my|*1|is)(i know your *1 is *2)::thanks for the info [keywordlistorder:i am*] CMD::SAVE::(i|am|*1)(%i)::i get it [keywordlistorder:*i will*] [keywordlistorder:*i will do is* CMD::SAVE::(i|will|*2)(quoting %u '%o')::atleast someone will then *2 ++++++++++++++++++++++++++++++++++++++++++++++++++
  DOAGAIN	When for a given input the control can be redirected to a different TAG in the data.dat file. ++++++++++++++++++++++++++++++++++++++++++++++++++ #--------------------------------------------------- #Forcing Control to reread the Data file using DOAGAIN Command #--------------------------------------------------- #[keywordvariable:\*] is a good option to elevate the relevance of [keyword] over other tags. #For example if input is "how to proceed to salat" then instead of answering from [keywordstart:how] #it will answer from this section. Note that if [keyword:salat] was used instead of [keywordvariable] #then it would have obtained a lower relevance than the [keywordstart] #NOTE: Any more options placed with \* will just be of no use. [keywordvariable:\*] data=data:kw success=response:kwbasic failure=response:kwfail [data:kw] #leave the Right hand side EMPTY if you want to avoid the topic salat=goto reference salat zakat=goto reference zakat resurrection= sex= [response:kwbasic] Search in some books. %v. we better stick to our main topic. %v %u, i don't know much about this. %v [response:kwfail] CMD::DOAGAIN::%i::[askformoreinfo] CMD::DOAGAIN::%i::[dead] ++++++++++++++++++++++++++++++++++++++++++++++++++ The control will go to the tag [askformoreinfo] or [dead] if the keywordvariable does not find a correct match in the provided [data:kw]. The response would be picked from the [askformoreinfo] or [dead] tag at random.
  LOOKUPIN	When for a given input the control can be redirected to a different TAG in the data.dat file. ++++++++++++++++++++++++++++++++++++++++++++++++++ #--------------------------------------------------- # Forcing Control to collect response from a AMC file # using the LOOKUPIN Command #--------------------------------------------------- [dead] CMD::LOOKUPIN::sayings.amc|keywordsearch->sayings::people say '%c' i'm not sure i understand what %u is writing. what do you mean, %u? CMD::LOOKUPIN::sayings.amc|keywordsearch->sayings::%c is what I think +++++++++++++++++++++++++++++++++++++++++++++++++ The LOOKUPIN command would cause the control to open SAYINGS.AMC and pick a line available under the keywordsearch->sayings TAG. The criteria of selection of the line is based on the maximum keywords in the input that are available in the data available under the selected TAG.
  MATH	Under Development +++++++++++++++++++++++++++++++++++++++++++++++++ #--------------------------------------------------- # Solving Word Problems in Mathematics using MATH Command #--------------------------------------------------- [keywordlistorder:add*to*] CMD::MATH::(*1+*2)::*1 plus *2 is %c [keywordlistorder:divide*by*and add*to it] CMD::MATH::((*1/*2)+(*3))::i think it would result to %c +++++++++++++++++++++++++++++++++++++++++++++++++ The pattern in [KEYWORDLISTORDER] would let ALIMbot identify the numbers and operators involved in the problem. The MATH command would be used to convert the problem into the mathematic equation which would be processed and the result would be available in %c variable. Currently, the Mathematics Processor is not made and hence the %c variable would just return the equation. Any Volunteers to work on this????
  More commands to come... If you think of some please mail me.

   

   
• [PENDING] Conversion of the ALIMbot response obtained for solving the second task on the ALMbot Concept page.

ALIMbot Database

ALIMbot uses a database similar to that of ELIZA's BRAIN by Phillip Massyn incorporating few but significant changes. The altered database format is easy to understand. The memory/brain of ALIMbot utilizes the following files,

Important Notes

• The format of data.dat
• The format of AMC files
• Linking AMC files from data.dat

© Documentation ALIMbot by Mohammed Faizul Wahid