Text Classification The Nave Bayes algorithm Adapted from

Text Classification : The Naïve Bayes algorithm Adapted from Lectures by Prabhakar Raghavan (Yahoo and Stanford) and Christopher Manning (Stanford) Prasad L 13 Naive. Bayes. Classify 1

Relevance feedback revisited n n n In relevance feedback, the user marks a number of documents as relevant/nonrelevant, that can be used to improve search results. Suppose we just tried to learn a filter for nonrelevant documents This is an instance of a text classification problem: n n n Two “classes”: relevant, nonrelevant For each document, decide whether it is relevant or nonrelevant The notion of classification is very general and has many applications within and beyond IR. Prasad L 13 Naive. Bayes. Classify 2

Standing queries n The path from information retrieval to text classification: n You have an information need, say: n n You want to rerun an appropriate query periodically to find news items on this topic n n Unrest in the Niger delta region I. e. , it’s classification not ranking Such queries are called standing queries n n Prasad Long used by “information professionals” A modern mass instantiation is Google Alerts L 13 Naive. Bayes. Classify 3 13. 0

Spam filtering: Another text classification task From: "" <takworlld@hotmail. com> Subject: real estate is the only way. . . gem oalvgkay Anyone can buy real estate with no money down Stop paying rent TODAY ! There is no need to spend hundreds or even thousands for similar courses I am 22 years old and I have already purchased 6 properties using the methods outlined in this truly INCREDIBLE ebook. Change your life NOW ! ========================= Click Below to order: http: //www. wholesaledaily. com/sales/nmd. htm ========================= 4 13. 0

Text classification: Naïve Bayes Text Classification n Today: n Introduction to Text Classification n Also widely known as “text categorization”. n Probabilistic Language Models n Naïve Bayes text classification n Prasad Multinomial Bernoulli Feature Selection L 13 Naive. Bayes. Classify 5

Categorization/Classification n Given: n A description of an instance, x X, where X is the instance language or instance space. n n n Issue: how to represent text documents. A fixed set of classes: C = {c 1, c 2, …, c. J} Determine: n The category of x: c(x) C, where c(x) is a classification function whose domain is X and whose range is C. n Prasad We want to know how to build classification functions (“classifiers”). L 13 Naive. Bayes. Classify 6 13. 1

Document Classification “planning language proof intelligence” Test Data: (AI) (Programming) (HCI) Classes: ML Training Data: Prasad learning intelligence algorithm reinforcement network. . . Planning Semantics Garb. Coll. planning temporal reasoning plan language. . . programming semantics language proof. . . Multimedia garbage. . . collection memory optimization region. . . (Note: in real life there is often a hierarchy, not present in the above problem statement; and also, you get papers on ML approaches to Garb. Coll. ) GUI. . . 7 13. 1

More Text Classification Examples: Many search engine functionalities use classification Assign labels to each document or web-page: n Labels are most often topics such as Yahoo-categories e. g. , "finance, " "sports, " "news>world>asia>business" n Labels may be genres e. g. , "editorials" "movie-reviews" "news“ n Labels may be opinion on a person/product e. g. , “like”, “hate”, “neutral” n Labels may be domain-specific e. g. , "interesting-to-me" : "not-interesting-to-me” e. g. , “contains adult language” : “doesn’t” e. g. , language identification: English, French, Chinese, … e. g. , search vertical: about Linux versus not e. g. , “link spam” : “not link spam” Prasad L 13 Naive. Bayes. Classify 8

Classification Methods (1) n Manual classification n n Used by Yahoo! (originally; now downplayed), Looksmart, about. com, ODP, Pub. Med Very accurate when job is done by experts Consistent when the problem size and team is small Difficult and expensive to scale n Prasad Means we need automatic classification methods for big problems L 13 Naive. Bayes. Classify 9 13. 0

Classification Methods (2) n Automatic document classification n Hand-coded rule-based systems n n Used by CS dept’s spam filter, Reuters, CIA, etc. Companies (Verity) provide “IDE” for writing such rules n n Standing queries: Commercial systems have complex query languages (everything in IR query languages + accumulators) n n Prasad E. g. , assign category if document contains a given boolean combination of words Accuracy is often very high if a rule has been carefully refined over time by a subject expert Building and maintaining these rules is expensive L 13 Naive. Bayes. Classify 10 13. 0

A Verity topic (a complex classification rule) n Note: n n maintenance issues (author, etc. ) Hand-weighting of terms 11

Classification Methods (3) n Supervised learning of a document-label assignment function n Many systems partly rely on machine learning (Autonomy, MSN, Verity, Enkata, Yahoo!, …) n n n n k-Nearest Neighbors (simple, powerful) Naive Bayes (simple, common method) Support-vector machines (new, more powerful) … plus many other methods No free lunch: requires hand-classified training data But data can be built up (and refined) by amateurs Note that many commercial systems use a mixture of methods Prasad L 13 Naive. Bayes. Classify 12

Probabilistic relevance feedback n n Rather than re-weighting in a vector space… If user has told us some relevant and some irrelevant documents, then we can proceed to build a probabilistic classifier, such as the Naive Bayes model we will look at today: n n P(tk|R) = |Drk| / |Dr| P(tk|NR) = |Dnrk| / |Dnr| n Prasad tk is a term; Dr is the set of known relevant documents; Drk is the subset that contain tk; Dnr is the set of known irrelevant documents; Dnrk is the subset that contain tk. L 13 Naive. Bayes. Classify 13 9. 1. 2

Recall a few probability basics n n For events a and b: Bayes’ Rule Prior Posterior n Odds: Prasad 14 13. 2

Bayesian Methods n Learning and classification methods based on probability theory. n n Bayes theorem plays a critical role in probabilistic learning and classification. Build a generative model that approximates how data is produced. Uses prior probability of each category given no information about an item. Categorization produces a posterior probability distribution over the possible categories given a description of an item (and prior probabilities). Prasad L 13 Naive. Bayes. Classify 15 13. 2

Bayes’ Rule Prasad L 13 Naive. Bayes. Classify 16

Naive Bayes Classifiers Task: Classify a new instance D based on a tuple of attribute values into one of the classes cj C MAP = Maximum Aposteriori Probability Prasad L 13 Naive. Bayes. Classify 17

Naïve Bayes Classifier: Naïve Bayes Assumption n P(cj) n n Can be estimated from the frequency of classes in the training examples. P(x 1, x 2, …, xn|cj) n n O(|X|n • |C|) parameters Could only be estimated if a very, very large number of training examples was available. Naïve Bayes Conditional Independence Assumption: n Assume that the probability of observing the conjunction of attributes is equal to the product of the individual probabilities P(xi|cj). Prasad L 13 Naive. Bayes. Classify 18

The Naïve Bayes Classifier Flu X 1 runnynose n n Prasad X 2 sinus X 3 cough X 4 fever X 5 muscle-ache Conditional Independence Assumption: Features (term presence) are independent of each other given the class: This model is appropriate for binary variables n Multivariate Bernoulli model 19 13. 3

Learning the Model C X 1 n X 3 X 4 X 5 X 6 First attempt: maximum likelihood estimates n Prasad X 2 simply use the frequencies in the data 20 13. 3

Problem with Max Likelihood Flu X 1 runnynose n n X 2 sinus X 3 cough X 4 fever X 5 muscle-ache What if we have seen no training cases where patient had no flu and muscle aches? Zero probabilities cannot be conditioned away, no matter the other evidence! Prasad 21 13. 3

Smoothing to Avoid Overfitting # of values of Xi Prasad L 13 Naive. Bayes. Classify 22

Smoothing to Avoid Overfitting # of values of Xi n Somewhat more subtle version Prasad L 13 Naive. Bayes. Classify overall fraction in data where Xi=xi, k extent of 23 “smoothing”

Stochastic Language Models n Models probability of generating strings (each word in turn) in the language (commonly all strings over ∑). E. g. , unigram model M 0. 2 the 0. 1 a 0. 01 man 0. 01 woman 0. 03 said 0. 02 likes …Prasad the man likes the woman 0. 2 0. 01 0. 02 0. 01 multiply L 13 Naive. Bayes. Classify P(s | M) = 0. 00000008 13. 2. 1

Stochastic Language Models n Model probability of generating any string Model M 1 0. 2 0. 01 Model M 2 the 0. 2 the class 0. 0001 sayst 0. 03 0. 0001 pleaseth 0. 02 0. 0001 yon 0. 1 0. 2 pleaseth 0. 2 yon 0. 0005 maiden 0. 01 0. 0001 woman sayst the class pleaseth 0. 01 0. 0001 0. 02 yon maiden 0. 0001 0. 0005 0. 1 0. 01 P(s|M 2) > P(s|M 1) 13. 2. 1

Unigram and higher-order models P( n n ) = P( ) P( | Unigram Language Models P( ) P( | ) Easy. Effective! ) n Bigram (generally, n-gram) Language Models n P( ) P( | ) P( Other Language Models n ) P( | ) Grammar-based models (PCFGs), etc. n Prasad | Probably not the first thing to try in IR L 13 Naive. Bayes. Classify 26 13. 2. 1

Using Multinomial Naive Bayes Classifiers to Classify Text: Basic method n n n Attributes are text positions, values are words. Still too many possibilities Assume that classification is independent of the positions of the words n Use same parameters for each position n Result is bag of words model (over tokens not types) Prasad L 13 Naive. Bayes. Classify 28

Naïve Bayes: Learning Algorithm n n From training corpus, extract Vocabulary Calculate required P(cj) and P(xk | cj) terms n For each cj in C do n docsj subset of documents for which the target class is cj n Textj single document containing all docsj n for each word xk in Vocabulary n nk number of occurrences of xk in Textj n n Prasad L 13 Naive. Bayes. Classify 29

Naïve Bayes: Classifying n n positions all word positions in current document which contain tokens found in Vocabulary Return c. NB, where Prasad L 13 Naive. Bayes. Classify 30

Naive Bayes: Time Complexity n Training Time: O(|D|Ld + |C||V|)) where Ld is the average length of a document in D. n n n Assumes V and all Di , ni, and nij pre-computed in O(|D|Ld) time during one pass through all of the data. Generally just O(|D|Ld) since usually |C||V| < |D|Ld Why? Test Time: O(|C| Lt) where Lt is the average length of a test document. n n Prasad Very efficient overall, linearly proportional to the time needed to just read in all the data. Plus, robust in practice L 13 Naive. Bayes. Classify 31

Exercise §Estimate parameters of Naive Bayes classifier §Classify test document 32 32

Example: Parameter estimates The denominators are (8 + 6) and (3 + 6) because the lengths of textc and are 8 and 3, respectively, and because the constant B is 6 as the vocabulary consists of six terms. 33 33

Example: Classification Thus, the classifier assigns the test document to c = China. The reason for this classification decision is that the three occurrences of the positive indicator CHINESE in d 5 outweigh the occurrences 34 of the two negative indicators JAPAN and TOKYO. 34

Underflow Prevention: log space n n Multiplying lots of probabilities, which are between 0 and 1, can result in floating-point underflow. Since log(xy) = log(x) + log(y), it is better to perform all computations by summing logs of probabilities rather than multiplying probabilities. Class with highest final un-normalized log probability score is still the most probable. Note that model is now just max of sum of weights… Prasad L 13 Naive. Bayes. Classify 35

Note: Two Models n Model 1: Multivariate Bernoulli n n n One feature Xw for each word in dictionary Xw = true in document d if w appears in d Naive Bayes assumption: n n Given the document’s topic, appearance of one word in the document tells us nothing about chances that another word appears This is the model used in the binary independence model in classic probabilistic relevance feedback in hand-classified data Prasad L 13 Naive. Bayes. Classify 36

Two Models n Model 2: Multinomial = Class conditional unigram n One feature Xi for each word pos in document n n n Value of Xi is the word in position i Naïve Bayes assumption: n n feature’s values are all words in dictionary Given the document’s topic, word in one position in the document tells us nothing about words in other positions Second assumption: n Word appearance does not depend on position for all positions i, j, word w, and class c Prasad 37

Parameter estimation n Multivariate Bernoulli model: fraction of documents of topic cj in which word w appears n Multinomial model: fraction of times in which word w appears across all documents of topic cj n n Can create a mega-document for topic j by concatenating all documents on this topic Use frequency of w in mega-document Prasad L 13 Naive. Bayes. Classify 38

Classification n Multinomial vs Multivariate Bernoulli? Multinomial model is almost always more effective in text applications! See IIR sections 13. 2 and 13. 3 for worked examples with each model Prasad L 13 Naive. Bayes. Classify 39

Feature Selection: Why? n Text collections have a large number of features n n 10, 000 – 1, 000 unique words … and more Feature Selection n Makes using a particular classifier feasible n n Reduces training time n n Training time for some methods is quadratic or worse in the number of features Can improve generalization (performance) n n Prasad Some classifiers can’t deal with 100, 000 of features Eliminates noise features Avoids overfitting L 13 Naive. Bayes. Classify 40 13. 5

Feature selection: how? n Two ideas: n Hypothesis testing statistics: n n n Information theory: n n n Are we confident that the value of one categorical variable is associated with the value of another Chi-square test ( 2) How much information does the value of one categorical variable give you about the value of another Mutual information (MI) They’re similar, but 2 measures confidence in association, (based on available statistics), while MI measures extent of association (assuming perfect knowledge of probabilities) Prasad L 13 Naive. Bayes. Classify 41 13. 5

2 statistic (CHI) n n n 2 is interested in (fo – fe)2/fe summed over all table entries: is the observed number what you’d expect given the marginals? The null hypothesis is rejected with confidence. 999, since 12. 9 > 10. 83 (the value for. 999 confidence). Term = jaguar Class = auto Class auto 2 (0. 25) 3 (4. 75) Term jaguar 500 expected: fe (502) 9500 (9498) observed: fo 42 13. 5. 2

2 statistic (CHI) There is a simpler formula for 2 x 2 2: A = #(t, c) B = #(t, ¬c) C = #(¬t, c) D = #(¬t, ¬c) Yields 1. 29? N=A+B+C+D Value for complete independence of term and category?

Feature selection via Mutual Information n n In training set, choose k words which best discriminate (give most info. on) the categories. The Mutual Information between a word, class is: n Prasad For each word w and each category c L 13 Naive. Bayes. Classify 44 13. 5. 1

Feature selection via MI (contd. ) n n For each category we build a list of k most discriminating terms. For example (on 20 Newsgroups): n n n sci. electronics: circuit, voltage, amp, ground, copy, battery, electronics, cooling, … rec. autos: car, cars, engine, ford, dealer, mustang, oil, collision, autos, tires, toyota, … Greedy: does not account for correlations between terms Prasad L 13 Naive. Bayes. Classify 45

Feature Selection n Mutual Information n Chi-square n n n Clear information-theoretic interpretation May select rare uninformative terms Statistical foundation May select very slightly informative frequent terms that are not very useful for classification Just use the commonest terms? n n Prasad No particular foundation In practice, this is often 90% as good L 13 Naive. Bayes. Classify 46

Feature selection for NB n n n In general, feature selection is necessary for multivariate Bernoulli NB. Otherwise, you suffer from noise, multi-counting “Feature selection” really means something different for multinomial NB. It means dictionary truncation. n Prasad This “feature selection” normally isn’t needed for multinomial NB, but may help a fraction with quantities that are badly estimated L 13 Naive. Bayes. Classify 47

Web. KB Experiment (1998) n Classify webpages from CS departments into: n n student, faculty, course, project Train on ~5, 000 hand-labeled web pages n Cornell, Washington, U. Texas, Wisconsin n Crawl and classify a new site (CMU) n Results:

NB Model Comparison: Web. KB Prasad 50

Prasad L 13 Naive. Bayes. Classify 51

Naïve Bayes on spam email Prasad L 13 Naive. Bayes. Classify 52 13. 6

Violation of NB Assumptions n n n Conditional independence “Positional independence” Examples? Prasad L 13 Naive. Bayes. Classify 54

Naive Bayes is Not So Naive n Naïve Bayes: First and Second place in KDD-CUP 97 competition, among 16 (then) state of the art algorithms Goal: Financial services industry direct mail response prediction model: Predict if the recipient of mail will actually respond to the advertisement – 750, 000 records. n Robust to Irrelevant Features cancel each other without affecting results Instead Decision Trees can heavily suffer from this. n Very good in domains with many equally important features Decision Trees suffer from fragmentation in such cases – especially if little data n n A good dependable baseline for text classification (but not the best)! Optimal if the Independence Assumptions hold: If assumed independence is correct, then it is the Bayes Optimal Classifier for problem n Very Fast: Learning with one pass of counting over the data; testing linear in the number of attributes, and document collection size n Low Storage requirements Prasad L 13 Naive. Bayes. Classify 57