NAME AI::ExpertSystem::Advanced - Expert System with backward, forward and mixed algorithms DESCRIPTION Inspired in AI::ExpertSystem::Simple but with additional features: * Uses backward, forward and mixed algorithms. * Offers different views, so user can interact with the expert system via a terminal or with a friendly user interface. * The knowledge database can be stored in any format such as YAML, XML or databases. You just need to choose what driver to use and you are done. * Uses certainty factors. Attributes initial_facts A list/set of initial facts the algorithms start using. During the forward algorithm the task is to find a list of goals caused by these initial facts (the only data it has at that moment). Lets imagine your knowledge database is about symptoms and diseases. You need to find what diseases are caused by the symptoms of a patient, these first symptons are the initial facts. initial_facts_dict For making easier your job, AI::ExpertSystem::Advanced asks you only the id of the "initial_facts". Once you provide them then a dictinary is created. This "initial_facts_dict" dictionary basically provides a standard interface to get the sign of the facts. goals_to_check When doing the "backward()" algorithm it's needed to have at least one goal (aka hypothesis). This could be pretty similar to "initial_facts", with the difference that the initial facts are used more with the causes of the rules, and this one with the goals (usually one in a well defined knowledge database). From our example of symptoms and diseases lets imagine we have the hypothesis that a patient has flu, we don't know the symptoms it has, we want the expert system to keep asking us for them to make sure that our hypothesis is correct. goals_to_check_dict Very similar to "goals_to_check" (and indeed of "initial_facts_dict"). We want to make the job easier at the moment of assigning goals and based on this only the list of goals is needed, then a dictionary will be created with the data of "goals_to_check". inference_facts Inference facts are basically the core of an expert system. These are facts that are found and copied when the a set of facts (initial, inference or asked) match with the causes of a goal. "inference_facts" is a AI::ExpertSystem::Advanced::Dictionary, it will store the name of the fact, the rule that caused these facts to be copied to this dictionary, the sign and the algorithm that triggered the copy. knowledge_db The object reference of the knowledge database AI::ExpertSystem::Advanced is using. asked_facts During the "backward()" algorithm there will be cases when there's no clarity if a fact exists. In these cases the "backward()" will be asking the user (via automation or real questions) if a fact exists. Going back to the "initial_facts" example of symptoms and diseases. Imagine the algorithm is checking a rule, some of the facts of the rule make a match with the ones of "initial_facts" or "inference_facts" but some wont, for these *unsure* facts the "backward()" will ask the user if a symptom (a fact) exists. All these asked facts will be stored here. visited_rules Keeps a record of all the rules the algorithms have visited. verbose By default this is turned off. If you want to know what happens behind the scenes turn this on. Everything that needs to be debugged will be passed to the "debug()" method of your "viewer". viewer Is the object AI::ExpertSystem::Advanced will be using for printing what is happening and for interacting with the user (such as asking the "asked_facts"). This is practical if you want to use a viewer object that is not provided by AI::ExpertSystem::Advanced::Viewer::Factory. viewer_class Is the the class name of the "viewer". You can decide to use the viewers AI::ExpertSystem::Advanced::Viewer::Factory offers, in this case you can pass the object or only the name of your favorite viewer. found_factor In your knowledge database you can give different *weights* to the facts of each rule (eg to define what facts have more *priority*). During the "mixed()" algorithm it will be checking what causes are found in the "inference_facts" and in the "asked_facts" dictionaries, then the total number of matches (or total number of certainity factors of each fact) will be compared versus the value of this factor, if it's higher or equal then the rule will be triggered. You can read the documentation of the "mixed()" algorithm to know the two ways this factor can be used. shot_rules All the rules that are shot are stored here. This is a hash, the key of each item is the rule id while its value is the precision time when the rule was shot. The precision time is useful for knowing when a rule was shot and based on that you can know what steps it followed so you can compare (or reproduce) them. Constants * FACT_SIGN_NEGATIVE Used when a fact is negative, aka, a fact doesn't happen. * FACT_SIGN_POSITIVE Used for those facts that happen. * FACT_SIGN_UNSURE Used when there's no straight answer of a fact, eg, we don't know if an answer will change the result. Methods shoot($rule, $algorithm) Shoots the given rule. It will do the following verifications: * Each of the facts (causes) will be compared against the "initial_facts_dict" and "asked_facts" (in this order). * If an initial or asked fact matches with a cause but it's negative then all of its goals (usually only one by rule) will be copied to the "initial_facts_dict" and "inference_facts" with a negative sign, otherwise a positive sign will be used. * Will add the rule to the "shot_rules" hash. is_rule_shot($rule) Verifies if the given $rule has been shot. get_goals_by_rule($rule) Will ask the "knowledge_db" for the goals of the given $rule. A AI::ExpertSystem::Advanced::Dictionary will be returned. get_causes_by_rule($rule) Will ask the "knowledge_db" for the causes of the given $rule. A AI::ExpertSystem::Advanced::Dictionary will be returned. is_fact_negative($dict_name, $fact) Will check if the given $fact of the given dictionary ($dict_name) is negative. copy_to_inference_facts($facts, $sign, $algorithm, $rule) Copies the given $facts (a dictionary, usually goal(s) of a rule) to the "inference_facts" dictionary. All the given goals will be copied with the given $sign. Additionally it will add the given $algorithm and $rule to the inference facts. So later we can know how we got to a certain inference fact. compare_causes_with_facts($rule) Compares the causes of the given $rule with: * Initial facts * Inference facts * Asked facts It will be couting the matches of all of the above dictionaries, so for example if we have four causes, two make match with initial facts, other with inference and the remaining one with the asked facts, then it will evaluate to true since we have a match of the four causes. get_causes_match_factor($rule) Similar to "compare_causes_with_facts()" but with the difference that it will count the *match factor* of each matched cause and return the total of this weight. The match factor is used by the "mixed()" algorithm and is useful to know if a certain rule should be shoot or not. The *match factor* is calculated in two ways: * Will do a sum of the weight for each matched cause. Please note that if only one cause of a rule has a specified weight then the remaining causes will default to the total weight minus 1 and then divided with the total number of causes (matched or not) that don't have a weight. * If no weight is found with all the causes of the given rule, then the total number of matches will be divided by the total number of causes. is_goal_in_our_facts($goal) Checks if the given $goal is in: * The asked facts * The inference facts remove_last_visited_rule() Removes the last visited rule and return its number. visit_rule($rule) Adds the given $rule to the end of the "visited_rules". copy_to_goals_to_check($facts) Copies a list of facts (usually a list of causes of a rule) to . ask_about($fact) Uses "viewer" to ask the user for the existence of the given "fact". The valid answers are: + or "FACT_SIGN_POSITIVE" In case user knows of it. - or "FACT_SIGN_NEGATIVE" In case user doesn't knows of it. ~ or "FACT_SIGN_UNSURE" In case user doesn't have any clue about the given fact. get_rule_by_goal($goal) Looks in the "knowledge_db" for the rule that has the given goal. If a rule is found its number is returned, otherwise undef. forward() The forward chaining algorithm is one of the main methods used in Expert Systems. It starts with a set of variables (known as initial facts) and reads the available rules. It will be reading rule by rule and for each one it will compare its causes with the initial facts and with the inference facts. If all of these causes are in our facts then the rule will be shoot and all of its goals will be copied/converted to inference facts and will restart reading from the first rule. backward() The backward algorithm starts with a set of *assumed* goals (facts). It will start reading goal by goal. For each goal it will check if it exists in the "asked_facts" and "inference_facts". * If the goal exist then it will be removed from the dictionary, it will also verify if there are more visited rules to shoot. If there are still more visited rules to shoot then it will take the last one and remove it. Then this visited rule will be shoot. Once the rule is shoot it verifies if there are still still more goals to check, if this is the case then it starts reading from the first goal (at this time the "goals_to_check_dict" is reduced by 1. However if there are no more goals to check then it will finish, making the end of the algorithm. In case there are no more visited rules to shoot then it will finish making the end of the algorithm. * If the goal doesn't exist in the "asked_facts" or "inference_facts" then the goal will be searched in the list of goals of all the rules. In case it finds the rule that has the goal, this rule will be marked (added) to the list of visited rules ("visited_rules"). Also all of the causes of this rule will be added to the top of the "goals_to_check_dict". Once this is done it will start reading again all of the goals to check. If there's the case where the goal doesn't exist as a goal in our rules then it will ask the user (via "ask_about") for the existence of it. If user is not sure about it then the algorithm ends. mixed() As its name says, it's a mix of "forward()" and "backward()" algorithms, it requires to have at least one initial fact. The first thing it does is to run the "forward()" algorithm (hence the need of at least one initial fact). If the algorithm fails then the mixed algorithm also ends unsuccessfully. Once the first *run* of "forward()" algorithm happens it starts looking for any positive inference fact, if only one is found then this ends the algorithm with the assumption it knows what's happening. In case no positive inference fact was found then it will start reading the rules and creating a list of intuitive facts. For each rule it will get a *certainty factor* of its causes versus the "initial_facts_dict", "inference_facts" and "asked_facts". In case the certainity factor is greater or equal than "found_factor" then all of its goals will be copied to the intuitive facts (eg, read it as: it assumes the goals have something to do with our first initial facts). Once all the rules are read then it verifies for any intuitive fact, if no facts are found then it ends with the intuition, otherwise it will run the "backward()" algorithm for each one of these facts (eg, each fact will be converted to a goal). After each *run* of the "backward()" algorithm it will verify for any positive inference fact, if just one is found then the algorithm ends. At the end (if there are still no positive inference facts) it will run the "forward()" algorithm and restarts (by looking again for any positive inference fact). A good example to understand how this algorithm is useful is: imagine you are a doctor and know some of the symptoms and diseases of a patient. Then the algorithm will start looking for any additional disease you could be missing. Then once it ends looking for diseases it will check if we know what the disease is (by looking for the positive fact). If there's still no clue then it starts looking in *viceversa*, now knowing a list of possible diseases and also a list of symptoms. It repeats all the process until a positive *inference* fact is found. summary($return) The main purpose of any expert system is the ability to explain: what is happening, how it got to a result, what assumption it required to make, which facts it excluded and which used. This method will use the "viewer" (or return the result) in YAML format of all the rules that were shot. It will explain how it got to each one of the causes so a better explanation can be done by the "viewer". If $return is defined (eg, it got any parameter) then the result wont be passed to the "viewer", instead it will be returned as a string. SEE ALSO Take a look AI::ExpertSystem::Simple too. AUTHOR Pablo Fischer (pablo@pablo.com.mx). COPYRIGHT Copyright (C) 2010 by Pablo Fischer. This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.