Google Opinions Project Proposal

GOOGLE OPINIONS PROJECT PROPOSAL

Prepared for

Professor Betsy Schlobohm

Prepared by

David Urbina

Khairun-nisa Hassanali

Michael Fashola

Scott Larson

April 28, 2009

GOOGLE OPINIONS

Memorandum

DATE:March 25, 2009

TO:Scott Larson – Project Manager, Google Think Team

FROM:Marissa Mayer - VP Search Products and User Experience

SUBJECT:Update to Google Search Engine

Recent advances in Information and Communication technology have made our search engine developed in the early days of the World Wide Web in need of major reviews and update. We have seen increase in the number of blogs, twittering, social networking sites and other opinion-themed sites like yelp.com,jdpower.com, and rottentomatoes.com that signal a paradigm shift in content and distribution [1]. No longer are opinions two dimensional images that exist only in the mind of the conceiver but, through the instrumentality of the web, are now potent agents of change [2].

Google as a leader in launching feature-rich applications has set up a new group called Google Think to undertake the task of enhancing the Google Search Engine in order to accommodate this paradigm shift and develop a contemporary search engine that will provide decision-making solutions to consumers.

I am happy to invite your team to undertake this project with the responsibility of designing and developing a comprehensive search engine that will mine blogs, opinion editorials and provide themed search which will ultimately appeal to users.

We wish you the best in this assignment.

Cheers,

Marissa Mayer

Memorandum

DATE:March 26, 2009

TO:Marissa Mayer – Vice President, Search Products and User Experience

FROM:Scott Larson – Project Manager, Google Think Team

SUBJECT:Acceptance of Offer

Your memo empowering our group to create an update to the Google search engine was received with enthusiasm.

Our appointment to undertake this project reflects the high level of confidence you repose on us to develop a ground-breaking application that will re-launch Google as a leader in search engine development. Please rest assured we will approach this assignment with vigor.

My team will be conducting surveys and may request for information that will aid a successful completion of this project. This project will involve the use of products and technologies developed by other teams at Google. We would therefore need the cooperation of all these teams at Google. Kindly inform us if there is to be any change in scope of this project.

Cheers,

Scott Larson

Memorandum

DATE:April 10, 2009

TO:Marissa Mayer – Vice President, Search Products and User Experience

FROM:Scott Larson – Project Manager, Google Think Team

SUBJECT:Updates to Google Search Engine

We have conducted an assessment of the current search engine in line with your request, and proffered viable options in enhancing the Google search engine as we know it today.

We are in agreement with the need to update the Google search engine capabilities. While Google’s share of searches has increased year-on-year [3],it is still unable to meet one of the primary needs of today’s internet users: searching for opinions.

We propose Google Opinions, an extension of the Google Search Engine, which provides an end to end solution to users searching for opinions. Google Opinions will also provide users with advanced search capabilities that will enable them to further pin down the source of the opinions and display statistics on the opinions retrieved by the system.

The general objective of the proposed plan is to make Google more responsive to the evolving internet culture [1], [2] and launch new capabilities that will put Google at the frontline for years to come. A critical success factor is to have a cross-over appeal to Google, so that it can be of use to all strata of the society. This better positioning will lead to an improvement in revenue streams.

We hereby acknowledge with thanks, the kind assistance of Vinton Cerf and his team in providing market intelligence reports and new markets product themes. These came in handy while analyzing the cost and benefits of the proposed option.

Working on the project proposal has broadened our insights. We envisage this is just a step in a series of concerted efforts to make Google a leader in her field, and we will be glad to play a role in helping realize Google’s short and long term goals.

Cheers,

Scott Larson.

Table of Contents

Memorandum

Memorandum

Memorandum

Table of Contents

List of Illustrations

List of Tables

Executive Summary

Google Opinions

Introduction

Current Situation

Project Plan

The solution

Objectives

Major and Minor Steps

Deliverables and outcomes

Qualifications

The Google Think Group

The People

Costs and Benefits

Conclusion and Recommendations

Appendix A.Project Plan

Google Opinions System

Google Opinions Project Timeline

Google Opinions Main Page

Appendix B.Costs and Revenue

Costs

Revenue

Appendix C.Google Opinions Glossary

Appendix D.Financial Statements

Google Inc.: Income Statement

Google Inc.: Balance Sheet

Google Inc.: Cash Flow Statement

Appendix E.Resumes

References

List of Illustrations

Figure A1: Google Opinions System Diagram

Figure A2: Google Opinions Project Time Line

Figure A3: Google Opinions User Interface

Figure B1: Cost Profile

Figure B2: Phase-wise Cost

Figure B3: Cost per Module

Figure B4: Cost Benefit Analysis

List of Tables

Table 1: Major and Minor Steps

Table 2: Project Deliverables and Outcomes

Table 3: Software Cost

Table 4: Phase-wise Cost

Table 5: Cost per Module

Table 6: Income Statement

Table 7: Balance Sheet

Table 8: Cash Flow Statement

Executive Summary

Google needs to once again re-position herself as a leader in search engine development, and offer greater value to users in real value terms. Opinion-themed searches will enable us seize the initiative and open new vistas in previously uncharted territory.

We propose Google Opinions, an extension of the Google Search Engine, to provide opinion-themed searches and appeal to a broad spectrum of users with varied needs such as consumers, employers, students and businessmen [1] –[3]. We envisage Google Opinions being used to provide opinions in a recruitment decision, product ratings in a purchasing decision and other individual and corporate decision challenges arising from the current environment of a vast array of products and an information overload.

The Google Opinions project is proposed to start on June 1, 2009 and end on May 31, 2010. This project will involve a team of thirty highly qualified personnel, with extensive experience in information retrieval, sentiment mining and software development.

The financial and non-financial benefits from the Google Opinions project far outweigh the cost. The Google Opinions project will cost $10,909,357. The increase in revenue from Google Opinions is envisaged to be above $4,000,000 per annum.

Implementing the Google Opinions project will lead to publishing path breaking papers on sentiment mining. Further, our competitors such as Microsoft [4]– [6] are working on similar technology and therefore it is imperative that we get Google Opinions out into the market before our competitors.

1

1

Google Opinions

Introduction

A widely valued, but rarely provided service is that of opinions. Customers are always in need of advice about other interests, frequently referring to reviews written by professionals and other customers [1, 2]. However, while these opinions may be useful, they are far from exhaustive, and do not allow a view of the wide variety of views available.

Google Opinions attempts to solve this problem by providing a search engine designed specifically for retrieving opinions.

Current Situation

As of April 2009, search engines do not have the capability of performing opinion based searches. The World Wide Web is abounding with opinions on blogs, newspapers, review sites and social network sites [1] – [3]. Given time and effort, a user can use a standard search engine to research these websites and analyze the few opinions they find. However, the rate at which the information is collected, the quality of opinions retrieved in this manner is time consuming and insufficient for most consumer and business needs.

The cause of this problem is the lack of an efficient means to collect and retrieve opinions from websites [1]–[3]. The collection and analysis of the large quantity and variety of opinions available on the Web is beyond the scope of a single user’s practical and willing effort to retrieve the corresponding results. The Google Search Engine was not designed with a perspective of retrieving opinions from websites.

If this problem is not solved, both users and businesses will continue to suffer from a lack of online consensus and views on particular products, positions, and ideals [1]– [3], [7], [8]. Google’s major competitor Microsoft has filed patents on similar technology [4]–[6]. Google needs to bring out a system that solves this problem before its competitors giving it an edge in the market of opinion-based retrieval. Not doing so would lead to a loss of revenue as people will move to the competitors search engine. Further, there would other benefits that Google would miss out on such as publishing path breaking research papers that result from this project.

Project Plan

The solution

In order to meet the demand for opinion based searches [1]– [3], we propose to develop Google Opinions [7], [8], an extension of the Google Search Engine. Google Opinions will enable a user to perform opinion-based search retrieving both positive and negative opinions on a particular subject.

Google Opinions will also provide the user with advanced search features such as specifying the specific source of opinions and how negative or positive the opinions retrieved by Google Opinions should be. Further, using Thought Stats, users can view statistics and graphs on the opinions retrieved for a search subject. Thought Stats also includes i-Util, whichgives a measure of satisfactionderived from opinions on products and services. Please refer to Appendix A for a detailed explanation of the Google Opinions system along with the system diagram and proposed user interface.Please refer to Appendix C for an explanation of the terms used in the Google Opinions project.

Objectives

Google Opinions must meet the following objectives:

1. Provide an end to end solution that allows for users to:
2. Search for opinions on a subject [7], [8].
3. Display links to articles that contain opinions on these subjects.
4. Display representative sentences containing these opinions [7], [8].
5. .Provide a user interface that will provide for users to:
6. Provide the search words on which they want an opinion.
7. Provide for advanced search options that allow the user to:
8. Specify the polarity (degree of positivity or negativity of the opinions retrieved by Google Opinions [7], [8].
9. Specify the sources Google Opinions should use for retrieving opinions.
10. Specify the time frame within which the opinions are expressed.
11. Provide help on using Google Opinions.
12. Provide the user with statistics and graphs on the opinions retrieved by Google Opinions.
13. Display advertisements related to the search words.
14. Allow for easy integration with other Google components such as Google [9] and Google AdSense [10].

Major and Minor Steps

The table below gives the major and minor steps of the Google Opinions Project along with the timelines. The Google Opinions project is expected to start on June 1, 2009 and end on May, 31 2009. Please refer toFigure A2in Appendix A for the Google Opinions project timeline.

Major Step/Component / Minor Step / Expected Start Date / Expected Completion Date
Pre-Requirements / Testing and installing software / June 1, 2009 / June 30, 2009
User Interface / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / September 31, 2009
Unit Testing / October 1, 2009 / October 31, 2009
Go-Op-Crawler / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / October 31, 2009
Unit Testing / November 1, 2009 / November 30, 2009
Pre-Processor / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / September 30, 2009
Unit Testing / October 1, 2009 / October 31, 2009
Go-Top-Generator / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / October 31, 2009
Unit Testing / November 1, 2009 / November 30, 2009
Go-Op-Selector / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / October 31, 2009
Unit Testing / November 1, 2009 / November 30, 2009
Go-Summarizer / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / October 31, 2009
Unit Testing / November 1, 2009 / November 30, 2009
Thought Stats / Requirements / July 1, 2009 / July 31, 2009
Design / August 1, 2009 / August 31, 2009
Coding / September 1, 2009 / October 31, 2009
Unit Testing / November 1, 2009 / November 30, 2009
Integration / All components / December 1, 2009 / December 31, 2009
System Integration Testing [11] / Testing and bug fixing / January 1, 2010 / March 31, 2010
Product Quality Testing [12] / Testing and bug fixing / April 1, 2010 / May 31, 2010

Table 1:Major and Minor Steps

Deliverables and outcomes

Table 2gives the deliverables and outcomes for the Google Opinions Project. These deliverables are due once the component is completed. Please refer to Table 1 for the completion dates of each component.

Component / Deliverables / Outcome
User Interface / Source Code [13] / A user interface which enables users to search for opinions. The user interface will provide for advanced search options and a Help section.
Executable [14]
Deployment Manual [15]
User Manual
Software Requirements Specification [16]
Go-Op-Crawler / Source Code [13] / A crawler that collects opinion-laden web pages based on the search words.
Executable [14]
Deployment Manual [15]
Software Requirements Specification [16]
Unit Testing Set Kit
Preprocessor / Source Code [13] / A program that takes web pages collected by the Go-Op-Crawler and produces plain text.
Executable [14]
Deployment Manual [15]
Software Requirements Specification [16]
Unit Testing Set Kit
Go-Top-Generator / Source Code [13] / A module that takes as input text and gives as output the topics discussed in the text.
Executable [14]
Deployment Manual [15]
Software Requirements Specification [16]
Unit Testing Set Kit
Topic Aspect Database
Go-Op-Selector / Source Code / A module that takes as input a set of topics and text and gives as output opinion-laden sentences within the text on the given topics.
Executable
Deployment Manual
Software Requirements Specification
Unit Testing Set Kit
Taxonomic Sentiment Database / A module that takes a set of sentences as input and selects sentences that represent a summary of the input sentences.
Go-Summarizer / Source Code [13]
Executable [14]
Deployment Manual [15]
Software Requirements Specification [16]
Unit Testing Set Kit
Thought Stats / Source Code [13] / A module that generates statistics and graphs based on the data collected by other modules.
Executable [14]
Deployment Manual [15]
Software Requirement Specification [16]
Unit Testing Set Kit

Table 2:Project Deliverables and Outcomes

Other general deliverables this project will produce are:

• System Architecture Document [17]
• Integration Testing Kit

Google Opinions will be the final deliverable which will provide an end to end solution enabling users to search for opinions. Google Opinions will be ready for deployment on May 31, 2009.

Qualifications

The Google Think Group

The Google Think group was founded three months ago to harness the potential market in opinion extraction and mining. The Google Think team consists of software engineers who are experienced in software development, sentiment mining, information retrieval and web based applications. Please refer to Appendix E for the resumes of the Google Think team members.

The People

Scott Larson – Manager

Scott is close to completing his Master’s Degree in Software Engineering at the University of Texas at Dallas. He is experienced in leading groups, including time management, resource management, work distribution, and project integration.Scott will be managing the Google Opinions project.

Khairun-nisa Hassanali – Lead Researcher and Developer

Khairun-nisa is a highly educated Researcher of Natural Language Processing at the University of Texas at Dallas. She is currently pursuing a PhD in Computer Science with a research focus on Sentiment Mining. Khairun-nisa will be leading all research and development on the Google Opinions project.

David Urbina – Lead Architect

David has extensive experience as a Solutions Architect and Developer.He is a Graduate Student at the University of Texas at Dallas with a 4.0 GPA, majoring in Software Engineering. David will be leading the Requirements Engineering, architecture, and design of the project.

Michael Fashola – Lead Tester and Manager of Marketing and Finance

Michael is a Graduate Student at the University of Texas at Dallas. He is experienced in software testing and financial auditing. Michael will be leading the Testing and Validation phase of the Google Opinions project, as well as managing the finances and marketing of the Google Opinions project.

Costs and Benefits

Our market research indicated that while Google has maintained her market share despite growing competition [18]– [21], opportunities for growth have been limited. This is supported by financial reports for preceding four quarters as shown in Appendix E.A key ingredient to jump-start another era of prolific growth is being able to offer quick and efficientsearch for opinions[1]–[3]. Google Opinions will fulfill this need. None of our competitors such as Microsoft[22] and Yahoo [23] have this capability in their search engines.

As shown in Appendix B, the implementation of the Google Opinions project will cost an estimated $10,909,357 dollars. We believe that this investment in Google Opinions will enable Google to take over a niche part of the market. We expect annual revenue increases of over $4,000,000 from Google Opinion and expect to break even in three years of implementing this product. This increase in revenue will come from Google AdWords [9] and AdSense [10] as businesses would want to purchase keywords for the Google Opinions related to their activities. A person looking for a positive or negative review on a product or service may also be looking at purchasing the same. Therefore, based on the reviews they are most likely to click on the business that sells this product or service. Further, Thought Stats is also a useful tool for businesses to track people’s opinions on their products and services. We intend to commercialize Thought Stats four years from now as a way of giving businesses access to otherwise copyrighted materials with expected revenues thereby further boosting our growth prospects.