Scribd
Upload
14 views

Assignment 1

This document outlines the steps to design a simple document search engine: 1. Define goals, gather documents, and choose a programming language. 2. Create a program to read, index, and search documents. The index lists words and their locations to enable fast searching. 3. Build a search function to lookup user queries in the index and return matching documents. Then display, test, and refine the search engine before documenting and sharing it.

Uploaded by

Balawal Rasool
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
14 views

Assignment 1

This document outlines the steps to design a simple document search engine: 1. Define goals, gather documents, and choose a programming language. 2. Create a program to read, index, and search documents. The index lists words and their locations to enable fast searching. 3. Build a search function to lookup user queries in the index and return matching documents. Then display, test, and refine the search engine before documenting and sharing it.

Uploaded by

Balawal Rasool
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 23

Introducing Indexes

By
Dr Syed Khaldoon Khurshid
• In searching a word in a document that you
could get away with searching the simplest
possible way: reading through the whole
document from beginning to end, looking
for the right words.
• That’s not how the any search engine you
used worked, and it wouldn’t be remotely
possible for Google to work this way!
• But we will develop our understanding of
indexes by using this simple approach.
• There's "one weird trick" behind any search
engine: building and using an index.
• A book’s index allows you to access a book in an
"inside out" manner. Instead of scanning the book to
find references to a topic, the index is an alphabetical
listing of topics that explains where in the book that
topic appears.
• Even huge internet search engines containing many
trillions of documents ultimately rely on storing all that
information in an index that is stored in a computer.
• The most fundamental operation on an index is
looking up a word. If the index is built just like a
book’s index, then a word can be found by looking in
the middle of the index, and then (depending on the
word) looking at the word in the middle of the first
half or the middle of the second half.
• By always looking in the middle, you can look up a word
in an index containing trillions of web pages by just
checking a few dozen entries! That’s the magic of binary

search!
• In a book, an index is a static document, created one time for
one book. There’s no space in a book’s index to add new
topics. That won’t work for a search engine: the web is
changing constantly!
• A search engine’s index has to support one other critical
operation: adding a new word. You can think of an index
that supports addition as a huge organized wall full of sticky
notes. It’s always possible to add a new sticky

note.
• In further lecture, you’re going to assume
that you have an index that supports these
two critical operations:
• quickly looking up words and
• quickly inserting new words.
• Based on these two operations, you’ll learn
how to build a index for search engine.
Search Engine that efficiently supports
complex queries over trillions of
documents.
Building An Index
• But to get big, you'll start very, very small.
In this module, you'll build an index that lets
you search these three "documents":
•
This collection contains 3 documents, with a
total of ten words.
• Your goal is to create an easily searchable record of
where words appear. In addition to your
documents, you
have a pad of sticky
notes. Start with the
first document:
• Ignoring punctuation and capitalization, the first
step is to write the first word, “Hope" on a sticky
note, along with the document identifier, D1:

• Repeat this process with the second and


third words in the document, and your wall
will look like this:
• Your goal is to use these sticky notes to
quickly look up individual words. Which of
the following is a good way to organize the
notes to facilitate this? (Remember that
sticky notes can be moved if and when
necessary.)
• After putting the sticky notes in alphabetical
order, the wall looks like this:
• Some information has been lost in representing
the first document this way. Which of the
following alternative documents would result
in the same arrangements of sticky notes?
(Select all that apply.)
• Now you can start on the second
document.

• The first word is a word you've encountered


before: "hope”.
• You already have a sticky note for "hope" on the
wall. Instead of adding a new sticky note, you use
the same one and just add the document number
D2 to the end of it, like this.

• The next word in D2 is "against." You'll need a new sticky note for
that one:

• The last word in D2 is "hope" again. The sticky note for "hope"
already has D2 on it, so you have a choice to make! • You could either
write 'D2' on the sticky note again or not. If you choose not to write
down D2 a second time on the
"hope" sticky note, and if you do the same thing for future
repeated words, what information will be lost?
• Continue on to the third document.
• The first word is "springs” In context, this has a

different meaning from the one you've


encountered before: it's a noun describing a bouncy
piece of metal, not a verb meaning “originates or
arises.”
• An automated indexer cannot make that distinction (at
least at this stage). For the purposes of the index,
these are treated as the same word and given only one
sticky note.
• This is really how it must be. If you search
for "jaguar" in Google, the search engine
cannot possibly know whether you meant
the car or the wild cat, and it must therefore
find pages about both.
• (In reality, Google does have sophisticated
algorithms to use external information such
as you and other people's search histories to
try to predict what you are probably looking
for, but a discussion of that is outside the
scope of our current studies.)
• You've now built an index for a small
database. The sticky notes and the wall are,
of course, a metaphor for something more
sophisticated happening within a computer
system. The basic principles, however,
remain the same!
• An index is a way to keep records for each
word in the database so that individual
words can be easily looked up and the
record points to all the documents
containing the word.
An Algorithm for Indexing
• Informally, you've seen how to create a search engine
out of sticky notes. In this exploration, you're going
to take a closer look at how this process can be
turned into a precise algorithm.
• Remember that the data structures behind a search
engine's index will support two kinds of operations
quickly: lookup and insert. In our sticky note
metaphor, these two operations look like this:
• Lookup: find a sticky note with a certain word on the
wall, or determine that no sticky note exists. • Insert:
place a sticky note on the wall so that it can be found
later.
• There's a third operation you'll want to pay attention
to. When you have found a sticky note on the wall or
created a new one, you can write to it to record new
information.

Here is
an algorithm for building a search engine's index. For each
document in our database and for each word in that document, run
the following procedure
• Why count the number of lookups and inserts
performed by the index-building algorithm at all? •
One reason is that understanding the number of
operations an algorithm performs is a good way to
better understand the algorithm.
• Another reason is that the lookup, insert, and write
operations, while hopefully quite fast, take up most
of the time of creating an index. This is true whether
you're building the index with powerful computers
or with sticky notes on your wall.
• Counting the number of operations enables us to
ignore the specifics of how we're putting together
the search engine's index, while still understanding
important details about how much time it will take
to build the index.
Case Study I for IR:
Design a Simple Document
Search Engine
Design a Simple Document Search Engine
• Creating a simple document search engine for information
retrieval involves several steps. Here's a step-by-step approach:
• Step 1: Define Your Goals
Understand what you want your search engine to do. Do you
want to search for specific words in documents? Do you want to
find documents related to certain topics?

• Step 2: Gather Documents


Collect the documents you want to search through. These could
be text files, PDFs, or web pages. Make sure they are in one place,
like a folder on your computer.

• Step 3: Choose a Programming Language


Decide on a programming language you are comfortable with.
Python is a good choice for beginners.
• Step 4: Create a Program
Write a computer program in your chosen language. This
program will help you search through your documents.

• Step 5: Read and Index Documents


Teach your program to read the text in your documents. It needs
to understand what's inside them.
Create an index. This is like a list of words and where they
appear in each document. It helps your search engine find things
faster.

• Step 6: Build a Search Function


Write a search function in your program. This function takes a
user's query (the words they want to search for) and looks it up
in the index you created.
When a user searches for something, your program will find
matching documents.
• Step 7: Display Results
Show the search results to the user. You can display
a list of documents that match their query.

• Step 8: Test Your Search Engine


Try different searches to make sure your search
engine works correctly. Make improvements if
needed.

• Step 9: Make it User-Friendly


Add a simple user interface. It could be a web page
or a command-line interface where users can type
their queries.
• Step 10: Refine and Optimize
Keep improving your search engine. Make it faster
and more accurate.
You can add features like ranking results based on
relevance.

• Step 11: Document Your Work


Write down how your search engine works, so
others can understand and use it.

• Step 12: Share Your Search Engine


If you want others to use your search engine, make it
accessible to them. You can host it online or share
the program.
Note:
• Remember, this is a simple search engine,
and there are more advanced features you
can add as you become more comfortable
with the basics. Start small, learn, and build
on your project over time.

You might also like