artificial intelligence questions

[Mehdi124]

can anyone help me solve these problems? i need some hint about at least fourth of them i would be thankful especially i know cilu can give hints..

Exercise 1: Find the difference between "خِرَد پارسی" and "rationality" and express it in a page (pdf or html format).
Exercise 2: Write a program that gets a keyword and creates a tree with this root node for a document. Then, the program builds a tree with nodes related to their parent nodes. This way, you create a summary oriented to the keyword (you can use any programming language that you like).
Help: Here we assume that here is only one document that we want to build such tree upon it. This may be used for helping the user to specifies what he wants. A simple relation for finding child nodes is the nodes that comes with the parent node in the document. For example, if in the document, "artificial intelligence" has come together, if the parent node is "intelligence", then one of its child nodes can be "artificial".
Exercise 3: Here we assume a query is given and we want to see most related documents to be ordered for us. We consider that we have quite a few text documents.
Help: define a quite number of index terms, we may say n index terms and order them as a vector: (t1, t2,…, tn). Every document is represented by a weight vector (wd1, wd2,…,wdn) where wdi denotes the importance (presence) of index term ti in the current document. One method for assigning to the weights is the tf-idf which is term frequency-inverse document frequency and is calculated as follows: wdi=(fti/max fj) * log N/ni such that max fj returns no. of times the most frequent index term in the current document appears. N is the total number of documents. ni is the no. of documents that have index term ti. Moreover, fti is the no. of times that the index term ti appears in the current document. Every given query is also represented by a weight vector which weights the query based on the index terms : (wq1. wq2,…,wqn) such that wqi represents the weight of query for the index term ti. (simply, wqi is zero if ti exists in the query or it is one when there is at least on instant of the index term ti in the current query). To find the similarly between a query and a document we calculate (using dot- product) as follows Sim(wd, wt)=(wd1.wt1+…+wdn.wtn)/Ö(wd12+…+wdn2) . Ö(wt12+…+wtn2)
Exercise 4: We assume a query is given in a form of an image, and the system should list the most similar images in a folder of many different images.
Help: use the color similarity measures to define a distance measure between images. One technique is use of color histogram, and compare the histogram of the query image with other images. a Color histogram of an image is a function H(r,g,b) such that H(r,g,b)=no. of times that the RGB color (r,g,b) appears in the image/total number of pixels of the image.
Exercise 5: Simulate a traffic environment in which cars move in different directions. Cars are agents such that each car begins from a source and finishes in a destination. The car should be destroyed as it reaches its destination. you can use any agent model you like to implement. The most important thing is that cars don't collide with each other. A search strategy is needed to guide the car to its destination, use any search technique you prefer. here we don't care about the shapes of cars, even simple rectangles can be used for cars.
Exercise 6: Solve the n-queen problem with a search algorithm that you like to implement. In an n-queen problem, the task is to put n queens on a chessboard with size n*n such that no two queens attack each others. We must randomly put the queens on board. Then, the algorithm shows the step that the queens move toward the better states.
Help: solve this as a constraint satisfaction problem or use local search algorithms.

i could nt get them as i skipped all classe and made fraudulent actions by devious means at final exam