DOC PREVIEW
MIT 6 871 - How to Score Free Drinks From Guys

This preview shows page 1-2-19-20 out of 20 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 20 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 20 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 20 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 20 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 20 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

6.871 Spring 2005 Final Project Paper 5/12/2005 How to Score Free Drinks From Guys 1. Motivation and Task We were motivated by the people's behaviors and social interactions at bars. Specifically, we noticed that certain women always have free drinks lined up, certain women always end up meeting and conversing with new people, certain men always comment on how short someone's skirt is, different men respond differently to various flirting styles, and that behaviors and dress can vary depending on the location. Intrigued by these patterned observations and reasoning behind human interactions at bars, we decided to capture that knowledge in our final project. We propose to build an advisor expert system that will help women get free drinks at bars. When the user, a woman, walks into a night scene, she will target a prospective man from whom she wishes to receive a drink. She will use our system by inquiring it for recommended course of actions. The system will, in response, pose a series of customized questions to the woman to collect information about the woman’s physical attributes and profession, the prospective man's behaviors and state, and the venue. Based on the inputs, the system will make several levels of inferences to make some conclusions about the woman's attractiveness, personality traits, the man's traits, and the setting characteristics. Using these intermediate inferences, the system will eventually result in a recommend course of action for the woman in her specific scenario. An example of user-system dialogue might look like this: (ask [action ann ?x] #'suggest)What is ANN's age: 40What is ANN's height: 61What is ANN's weight: 150Is it the case that ANN wears glasses: YesIs it the case that ANN has braces: No Is it the case that ANN has blue eyes: YesIs it the case that ANN has blonde hair: Yes What is ANN's hair length: LongWhat is ANN's shoe type: HeelsWhat is ANN's clothing fit: LooseIs it the case that ANN is wearing a skirt: NoWhat is ANN's profession: EducationWhat is ANN's location(venue type): ClubIs it the case that ANN's target man is wearing a suit: YesIs it the case that ANN's target man is alone: YesIs it the case that ANN's target man is speaking loudly: YesIs it the case that ANN's target man is making large gestures: Yes[ACTION ANN ASK-THE-BARTENDER-FOR-A-MENU-AND-KNOCK-OVER-YOUR-DRINK-WHEN-YOU-REACH-FOR-IT] 1.0[ACTION ANN PLACE-YOUR-GLASS-ON-THE-EDGE-OF-THE-BAR-NEXT-TO-HIS-ELBOW-AND-WAIT-UNTIL-YOUR-DRINK-GOES-FLYING-OFF] 0.64000005 In this example, the user, Ann, is not at an optimal age or weight for her height. However, she does have working for her the blue eyes, blond hair, and heels. So overall she is considered average in terms of attractiveness. Her profession in education indicates that she is experienced in working with people and is likely to be extroverted. Additionally, as an educator who wears loose clothing to go out at night, she is probably also reserved and not really creative. The man she is targeting is well dressed so he is probably respectable and wealthy. From the observation that he is talking loudly and gesturing wildly, one might infer that he is drunk as well. Given these intermediate conclusions, an attractive or daring woman would be able to easily take advantage of the prospective man's lack of company and his qualities. However, since Ann is reserved and only averagely attractive, she does not have as many options, and the approaches that will give her the greatest chance of success would take advantage of the conscience of the bartender or the man by tricking one into thinking he has spilled her drink. This example characterizes the type of problems the free drinks advisor system can handle. Our system is capable of taking objective information that the user should know or be able to observe and through the process of elimination and inferences, find successful paths to viable solutions. One key point to make is that the scope of our problem is recommending plausible approaches to receiving a free drink, given the relevant input information about the woman, man, and venue. The recommendations are useful given that the user has inputted correct information without fabricating any answers, such as her weight, and that it is her first time meeting the relevant parties involved in the scenario. The problems of trying to get several drinks from one man or from a man she already knows are both out of scope for the functionalities of our system. An obvious example is that if a woman succeeds the first time by stealing a man's drink, reasonably if he noticed, the second time he would not allow that to happen again. This type of problem only differs from the problem in scope slightly, but due to the volatile nature of human thoughts on which the success of our advisor system depends, it becomes out of scope. 2. Knowledge Representation We have chosen a rule-based knowledge representation since it seems to best match the scope of the problem we aimed to solve. In particular, rules provide the advantages of an easily extensible knowledge base, modularity of chunks of knowledge, and the ease of categorization of associations. Our knowledge base includes both factual knowledge and heuristic knowledge. An example rule of factual knowledge is the calculation of the Body Mass Index from the user's height and weight, and an example rule of heuristic knowledge is the judgmental assumption that a woman working in the design industry is extroverted. Given the scope of our problem, which involves an immense amount of variables and perceptions, it is no surprise that an overwhelming majority of the knowledge is heuristic. This in turn meansthat the knowledge base has much potential to be expanded in order to account for more of the variables and to capture more precise knowledge. Thus, it is important that our knowledge representation permits easy ways of expanding and modifying the knowledge base. As we learned in class, rule’s independent nature provides this flexibility. Next, as briefly discussed in the task description, the knowledge of the system captures two main types of knowledge: (1) association between observable attributes and intermediate qualities such as attractiveness; and (2) association between intermediate qualities and recommendations. An example rule that represent type (1) dictates that if the woman wears tight clothing or if her profession is administrative, marketing, education,


View Full Document

MIT 6 871 - How to Score Free Drinks From Guys

Download How to Score Free Drinks From Guys
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view How to Score Free Drinks From Guys and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view How to Score Free Drinks From Guys 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?