Programming Assignment 4

Little Search Engine

In this assignment you will implement a simple search engine for text documents using hash tables.

Worth 75 points = 7.5% of your course grade

Posted Fri, Nov 3

Due Wed, Nov 22, 11:00 PM (WARNING!! NO GRACE PERIOD)

Extended deadline (with ONE time free extension pass): Mon, Nov 27, 11:00 PM (NO GRACE PERIOD)

You get ONE free extension pass for assignments during the semester, no questions asked. There will be a total of 5 assignments this semester, and you may use this one free extension pass for any of the 5 assignments.

A separate Sakai assignment will be opened for extensions AFTER the deadline for the regular submission has passed. The regular submission deadline for all assignments will be on a Friday, 11 PM, and the deadline for the corresponding extensions will be on the following Monday, 11 PM.

You will work on this assignment individually. Read DCS Academic Integrity Policy for Programming Assignments - you are responsible for abiding by the policy. In particular, note that "All Violations of the Academic Integrity Policy will be reported by the instructor to the appropriate Dean".

IMPORTANT - READ THE FOLLOWING CAREFULLY!!!

Assignments emailed to the instructor or TAs will be ignored--they will NOT be accepted for grading. We will only grade submissions in Sakai.

If your program does not compile, you will not get any credit.

Most compilation errors occur for two reasons:

You are programming outside Eclipse, and you delete the "package" statement at the top of the file. If you do this, you are changing the program structure, and it will not compile when we test it.

You make some last minute changes, and submit without compiling.

To avoid these issues, (a) START EARLY, and give yourself plenty of time to work through the assignment, and (b) Submit a version well before the deadline so there is at least something in Sakai for us to grade. And you can keep submitting later versions (up to 10) - we will accept the LATEST version.

Summary

You will implement a little search engine to do two things: (a) gather and index keywords that appear in a set of plain text documents, and (b) search for user-input keywords against the index and return a list of matching documents in which these keywords occur.

Implementation

Download the attached lse_project.zip file to your computer. DO NOT unzip it. Instead, follow the instructions on the Eclipse page under the section "Importing a Zipped Project into Eclipse" to get the entire project, called Little Search Engine, into your Eclipse workspace.

Here are the contents of the project:

A class, lse.LittleSearchEngine. This is where you will fill in your code, details follow.

A supporting class, lse.Occurrence, which you will NOT change.

Two sample text documents, AliceCh1.txt, and WowCh1.txt, directly under the project folder, for preliminary testing. Be sure to get other online text documents--or make your own--for more rigorous testing.

A noisewords.txt file that contains a list of "noise" words, one per line. Noise words are commonplace words (such as "the") that must be ignored by the search engine. You will use this file (and this file ONLY) to filter out noise words from the documents you read, when gathering keywords.

A docs.txt file that has a list of all documents (in this case AliceCh1.txt and WowCh1.txt) from which the search engine should extract keywords.

NOTE: You will need to write your own driver to test your implementation. This driver can take as inputs a file that contains the names of all the documents (such as docs.txt), as well as the noisewords.txt file. It can then set up a LittleSearchEngine object and call its methods as needed to test the implementation. The docs.txt and noisewords.txt filenames will be sent in as the arguments to the makeIndex method in LittleSearchEngine.

Following is the sequence of method calls that will be performed on a LittleSearchEngine object, to index and search keywords.

LittleSearchEngine() - Already implemented.

The constructor creates new (empty) keywordsIndex and noiseWords hash tables. The keywordsIndex hash table is the MASTER hash table, which indexes all keywords from all input documents. The noiseWordshash table stores all the noise words. Both of these are fields in the LittleSearchEngine class.

Every key in the keywordsIndex hash table is a keyword. The associated value for a keyword is an array list of (document,frequency) pairs for the documents in which the keyword occurs, arranged in descending order of frequencies. A (document,frequency) pair is held in an Occurrence object. The Occurrence class is defined in the LittleSearchEngine.java file, at the top. In an Occurrence object, the document field is the name of the document, which is basically the file name, e.g. AliceCh1.txt.

void makeIndex(String docsFile, String noiseWordsFile) - Already implemented.

Indexes all the keywords in all the input documents. See the method documentation and body in the LittleSearchEngine.java file for details.

If you want to index the given sample documents, the first parameter would be the file docs.txt and the second parameter would be the noise words file, noisewords.txt

After this method finishes executing, the full index of all keywords found in all input documents will be in the keywordsIndex hash table.

The makeIndex methods calls methods loadKeywordsFromDocument and mergeKeywords, both of which you need to implement.

HashMap loadKeywordsFromDocument(String docFile) - You implement.

This method creates a hash table for all keywords in a single given document. See the method documentation for details.

This method MUST call the getKeyword method, which you need to implement.

String getKeyword(String word) - You implement.

Given an input word read from a document, it checks if the word is a keyword, and returns the keyword equivalent if it is.

FIRST, see the method documentation in the code for details, including a specific short list of punctuations to consider for filtering out. THEN, look at the following illustrative examples of input word, and returned value.

Input Parameter	Returned value
distance.	distance (strip off period)
equi-distant	null (not all alphabetic characters)
Rabbit	rabbit (convert to lowercase)
Through	null (noise word)
we're	null (not all alphabetic characters)
World...	world (strip trailing periods)
World?!	world (strip trailing ? and !)
What,ever	null (not all alphabetic characters)

Observe that (as per the rules described in the method documentation), if there is more than one trailing punctuation (as in the "World..." and "World?!" examples above), the method strips all of them. Also, the last example makes it clear that punctuation appearing anywhere but at the end is not stripped, and the word is rejected.

Note that this is a much simplified filtering mechanism, and will reject certain words that might be accepted by a real-world engine. But the idea is to not unduly complicate this process, focusing instead on hash tables, which is the point of this assignment. So, just stick to the rules described here.

void mergeKeywords - You implement.

Merges the keywords loaded from a single document (in method loadKeywordsFromDocument) into the global keywordsIndex hash table.

See the method documentation for details. This method MUST call the insertLastOccurence method, which you need to implement.

ArrayList insertLastOccurrence(ArrayList occs) - You implement.

See the method documentation for details. Note that this method uses binary search on frequency values to do the insertion. The return value is the sequence of mid points encountered during the search, using the regular (not lazy) binary search we covered in class. This return value is not used by the calling method-it is only going to be used for grading this method.

For example, suppose the list had the following frequency values (including the last one, which is to be inserted):

 -------------------- 12 8 7 5 3 2 6 -------------------- 0 1 2 3 4 5 6 

Then, the binary search (on the list excluding the last item) would encounter the following sequence of midpoint indexes:

 2 4 3 

Note that if a subarray has an even number of items, then the midpoint is the last item in the first half.

After inserting 6, the input list would be updated to this:

 -------------------- 12 8 7 6 5 3 2 -------------------- 0 1 2 3 4 5 6 

and the sequence 2 4 3 would be returned.

If the new item is a duplicate of something that already exists, it doesn't matter if the new item is placed before or after the existing item.

Note that the items are in DESCENDING order, so the binary search would have to be done accordingly.

ArrayList top5search(String kw1, String kw2) - You implement.

This method computes the search result for the input "kw1 OR kw2", using the keywordsIndex hash table. The result is a list of NAMES of documents (limited to the top 5) in which either of the words "kw1" or "kw2" occurs, arranged in descending order of frequencies. See the method documentation in the code for additional details.

As an example, suppose the search is for "deep or world", in the given test documents, AliceCh1.txt (call it A) and WowCh1.txt (call it W). The word "deep" occurs twice in A and once in W, and the word "world" occurs once in A and 7 times in W:

 deep: (A,2), (W,1) world: (W,7), (A,1) 

The result of the search is:

 WowCh1.txt, AliceCh1.txt 

in that order. (Recall that the name of a document is the same as the name of the document file.)

NOTE:

If there are no matches for either keyword, return null or empty list, either is fine.

If a document occurs in both keywords' match list, consider the one with the higher frequency - do NOT add frequencies.

Return AT MOST 5 non-duplicate entries. This means if there are more than 5 non-duplicate entries, then return the five top frequency entries, but if there are fewer than 5 non-duplicate entries, then return all of them.

If a document in the first match list (for the first keyword) has the same frequency as a document in the second match list (for the second keyword), and both are candidates for inclusion in the output (they are not the same document), then pick the document in the first list before the document in the second list.

Implementation Rules

Do NOT change the package name on the first line.

Do NOT add any import statements. With the existing imports, you may use any of the classes in packages java.lang, java.io and java.util.

Do NOT change the Occurrence class in ANY way.

Do NOT change the headers of any of the existing methods in LittleSearchEngine in ANY way.

Do NOT change the code in any of the implemented methods in ANY way.

Do NOT add any class fields in LittleSearchEngine.

You MAY add helper methods to LittleSearchEngine, but you must make them private.

Grading

Method	Points
getKeyword	10
loadKeywordsFromDocument	15
insertLastOccurrence	10
mergeKeywords	20
top5search	20

When a method is graded, the correct versions of other methods will be used. Also, all data structures will be set to their correct (expected) states before a method is called.

You need not do any error checking in your program for bad inputs.

Submission

Submit your LittleSearchEngine.java file.

package lse;

import java.io.*;

import java.util.*;

/**

* This class builds an index of keywords. Each keyword maps to a set of pages in

* which it occurs, with frequency of occurrence in each page.

*

*/

public class LittleSearchEngine {

/**

* This is a hash table of all keywords. The key is the actual keyword, and the associated value is

* an array list of all occurrences of the keyword in documents. The array list is maintained in

* DESCENDING order of frequencies.

*/

HashMap> keywordsIndex;

/**

* The hash set of all noise words.

*/

HashSet noiseWords;

/**

* Creates the keyWordsIndex and noiseWords hash tables.

*/

public LittleSearchEngine() {

keywordsIndex = new HashMap>(1000,2.0f);

noiseWords = new HashSet(100,2.0f);

}

/**

* Scans a document, and loads all keywords found into a hash table of keyword occurrences

* in the document. Uses the getKeyWord method to separate keywords from other words.

*

* @param docFile Name of the document file to be scanned and loaded

* @return Hash table of keywords in the given document, each associated with an Occurrence object

* @throws FileNotFoundException If the document file is not found on disk

*/

public HashMap loadKeywordsFromDocument(String docFile)

throws FileNotFoundException {

/** COMPLETE THIS METHOD **/

// following line is a placeholder to make the program compile

// you should modify it as needed when you write your code

return null;

}

/**

* Merges the keywords for a single document into the master keywordsIndex

* hash table. For each keyword, its Occurrence in the current document

* must be inserted in the correct place (according to descending order of

* frequency) in the same keyword's Occurrence list in the master hash table.

* This is done by calling the insertLastOccurrence method.

*

* @param kws Keywords hash table for a document

*/

public void mergeKeywords(HashMap kws) {

/** COMPLETE THIS METHOD **/

}

/**

* Given a word, returns it as a keyword if it passes the keyword test,

* otherwise returns null. A keyword is any word that, after being stripped of any

* trailing punctuation, consists only of alphabetic letters, and is not

* a noise word. All words are treated in a case-INsensitive manner.

*

* Punctuation characters are the following: '.', ',', '?', ':', ';' and '!'

*

* @param word Candidate word

* @return Keyword (word without trailing punctuation, LOWER CASE)

*/

public String getKeyword(String word) {

/** COMPLETE THIS METHOD **/

// following line is a placeholder to make the program compile

// you should modify it as needed when you write your code

return null;

}

/**

* Inserts the last occurrence in the parameter list in the correct position in the

* list, based on ordering occurrences on descending frequencies. The elements

* 0..n-2 in the list are already in the correct order. Insertion is done by

* first finding the correct spot using binary search, then inserting at that spot.

*

* @param occs List of Occurrences

* @return Sequence of mid point indexes in the input list checked by the binary search process,

* null if the size of the input list is 1. This returned array list is only used to test

* your code - it is not used elsewhere in the program.

*/

public ArrayList insertLastOccurrence(ArrayList occs) {

/** COMPLETE THIS METHOD **/

// following line is a placeholder to make the program compile

// you should modify it as needed when you write your code

return null;

}

/**

* This method indexes all keywords found in all the input documents. When this

* method is done, the keywordsIndex hash table will be filled with all keywords,

* each of which is associated with an array list of Occurrence objects, arranged

* in decreasing frequencies of occurrence.

*

* @param docsFile Name of file that has a list of all the document file names, one name per line

* @param noiseWordsFile Name of file that has a list of noise words, one noise word per line

* @throws FileNotFoundException If there is a problem locating any of the input files on disk

*/

public void makeIndex(String docsFile, String noiseWordsFile)

throws FileNotFoundException {

// load noise words to hash table

Scanner sc = new Scanner(new File(noiseWordsFile));

while (sc.hasNext()) {

String word = sc.next();

noiseWords.add(word);

}

// index all keywords

sc = new Scanner(new File(docsFile));

while (sc.hasNext()) {

String docFile = sc.next();

HashMap kws = loadKeywordsFromDocument(docFile);

mergeKeywords(kws);

}

sc.close();

}

/**

* Search result for "kw1 or kw2". A document is in the result set if kw1 or kw2 occurs in that

* document. Result set is arranged in descending order of document frequencies. (Note that a

* matching document will only appear once in the result.) Ties in frequency values are broken

* in favor of the first keyword. (That is, if kw1 is in doc1 with frequency f1, and kw2 is in doc2

* also with the same frequency f1, then doc1 will take precedence over doc2 in the result.

* The result set is limited to 5 entries. If there are no matches at all, result is null.

*

* @param kw1 First keyword

* @param kw1 Second keyword

* @return List of documents in which either kw1 or kw2 occurs, arranged in descending order of

* frequencies. The result size is limited to 5 documents. If there are no matches, returns null.

*/

public ArrayList top5search(String kw1, String kw2) {

/** COMPLETE THIS METHOD **/

// following line is a placeholder to make the program compile

// you should modify it as needed when you write your code

return null;

}

}

package lse;

/**

* This class encapsulates an occurrences of a keyword in a document. It stores the

* document name, and the frequency of occurrence in that document. Occurrences are

* associated with keywords in an index hash table.

*

* @author RU-NB-CS112

*

*/

public class Occurrence {

/**

* Document in which a keyword occurs.

*/

String document;

/**

* The frequency (number of times) the keyword occurs in the above document.

*/

int frequency;

/**

* Initializes this occurrence with the given document,frequency pair.

*

* @param doc Document name

* @param freq Frequency

*/

public Occurrence(String doc, int freq) {

document = doc;

frequency = freq;

}

/* (non-Javadoc)

* @see java.lang.Object#toString()

*/

public String toString() {

return "(" + document + "," + frequency + ")";

}

}

//docs.textfile//

AliceCh1.txtWowCh1.txt

//noisewords.textfile//

about after all also an and another any are as at be because been before being between both but by came can come could did do does each else for from get got has had he have her here him himself his how if in into is it its just like make many me might more most much must my never now of on only or other our out over re said same see she should since so some still such take than that the their them then there these they this those through to too under up use very want was way we well were what when where which while who will with would you your a b c d e f g h i j k l m n o p q r s t u v w x y z