Loading...

Google Semantic Search in Ireland. Personalisation, Trust, Influence, Reliance, Reputation, and Ethics

An Irish Discussion

Master's Thesis 2015 103 Pages

Communications - Public Relations, Advertising, Marketing, Social Media

Excerpt

Table Of Contents

Acknowledgements

List Of Figures

List Of Tables

Abstract

Chapter 1: Introduction
1.1 Introduction to Study
1.2 Focus of Research
1.3 Structure of Thesis

Chapter 2: Literature Review
2.1 Introduction
2.2 The Evolution of Google Search keywords
2.3 Google Personalisation
2.4 Echo Chamber Effect
2.5 The Filter Bubble
2.6 The Knowledge Graph - things, not strings
2.7 Search Psychology
2.8 Reliance on the Internet as an external memory bank
2.9 User sentiment
2.10 Search Influence
2.11 Ethics
2.12 Google's Monopoly
2.13 Antitrust
2.14 Trust &Reputation
2.15 Conclusion

Chapter 3: Research Methodology
3.1 Introduction
3.2 Epistemology
3.3 Theoretical Perspective
3.4 Research Methodology
3.5 Research Methods
3.6 Quantitative Research
3.6.1 Online Survey
3.6.2 Data Collection - General Public Online Survey
3.6.3 Data Collection - Technical and Expert Group Surveys
3.6.4 Rationale for using online surveys
3.7 Qualitative Research
3.7.1 Interview
3.7.2 Data Collection - Technical Group Interviews
3.7.3 Rationale for using interviews
3.7.4 Questionnaires
3.7.5 Data Collection - Expert Conversational Questionnaire
3.7.6 Rationale for using Conversational Expert Questionnaire
3.8 Reliability and Validity
3.9 Ethics
3.10 Limitations
3.11 Conclusion

Chapter 4: Quantitative Research Results And Analysis
4.1 Introduction
4.2 Quantitative Results and Discussion
4.2.1 Survey Age Demographic
4.2.2 Google Personalisation Awareness
4.2.3 Google Personalisation Results Satisfaction
4.2.4 Google Influence Concerns
4.2.5 Why Use Google and would you consider changing Search Engine?
4.2.6 Personalised Advertising Influence - Remarketing
4.2.7 Ethics
4.3 Survey Conclusion

Chapter 5: Qualitative Research Results And Analysis
5.1 Introduction
5.2 Qualitative Results and Discussion - Technical Interviews
5.2.1 Google Personalisation
5.2.2 Personalisation Influence
5.2.3 Why Use Google?
5.2.4 Google Ads Remarketing/Retargeting Users
5.2.5 Google and Privacy
5.2.6 Is Personalisation Ethical?
5.3 Qualitative Results and Discussion - Expert Conversational Questionnaire
5.3.1 Survey Results Opinions
5.3.2 The Evolution of Semantic Search
5.3.3 Trust in Google
5.3.4 Reliance on Google
5.3.5 Personalisation and Ethics
5.3.6 The future of Google's Monopoly
5.4 Qualitative Conclusion

Chapter 6: Research Conclusion
6.1 Overview
6.2 Objectives
6.3 Recommendations
6.4 Self-Reflection

References

Appendices
Appendix A - Online Survey Questions
Appendix B - Online Survey participants
Appendix C - Technical Group Interview Sample

Acknowledgements

My decision to undertake this Masters degree was made with much trepidation. A very difficult year in the most personal of senses almost scuppered my attempts. I almost never wrote a word – The fantastic people around me through their belief in me, and their love and support enabled me to continue.

I wish to thank my academic supervisor, Kevin Griffin, for convincing me to dust myself down and prove I had this dissertation in me. His guidance was both timely and invaluable. Our conversations about Print were always enjoyable. His passion for learning was always encouraging and refreshing.

I also wish to thank my Industry Supervisor, Joanne Casey, for the vast knowledge, help and advice she shared with me. Her attention to detail and her passion for digital marketing is something I will always strive towards.

Thank you to my Mother Claire and Brothers Terry, Ronan and Sister Deirdre for encouraging me even when my confidence was low.

I wish to dedicate this dissertation and the last four years of study to my beautiful wife Jennifer who was always there for me especially during those moments of personal difficulty. She understood from the beginning the passion I had for my studies. She was with me every word of this Thesis and for that I am ever grateful.

Finally I wish to thank my late father Pierce and late brother John who both understood the power of education and the freedom that knowledge could give. I miss you both so much and I hope you are watching down on me and are proud of the work I have done. Thank you all.

List Of Figures

Figure 1 - The Evolution of Google Search

Figure 2 - Pariser's Filter Bubble

Figure 3 - Relationship between epistemology-theorectical perspective-methodology-methods

Figure 4 - Conceptual Framework

Figure 5 - Best day and times based on maximum engagement to post on Facebook

Figure 6 - General Online Survey Age Demographic Participation

Figure 7 - The research process and degree of structure

Figure 8 - Age Demographic Categories

Figure 9 - Personalisation Awareness

Figure 10 - Personalisation Satisfaction

Figure 11 - Concern over influence

Figure 12 - Why Use Google for Search?

Figure 13 - Consider Changing Search Engine?

Figure 14 - Remarketing Ad Awareness.

Figure 15 - Remarketing Ad Influence.

Figure 16 - Is Personalisation Ethical

List Of Tables

Table 1 - Attributes of Quantitative Research

Table 2 - Advantages and Disadvantages of Survey research

Table 3 - Attributes of Qualitative Research

Table 4 - Technical Interviews Overview conducted March - June 2015

Table 5 - Online Survey Overview

Table 6 - Technical Interviews Overview conducted March - June 2015

Table 7 - Expert Questionnaire Overview

Table 8 - Expert Participants Overview

Abstract

This research is an investigation into the Irish public's perception of Google search. The study examines the methods used by Google to present users with the most relevant and satisfactory search results. 95% of the online public in Ireland use Google as a search tool. This study aims to uncover true user sentiment in Ireland. The objectives of this research are to reveal how the Irish public view Google in regards to semantic search personalisation, trust, influence, reliance, reputation, and ethics.

The existing literature revealed that Google's search engine has evolved from, one that used Boolean Search logic to match keywords, to a semantic engine that understands user intent and context. The literature also reveals how the younger 'Millennial' generation associate the Internet with one brand, Google. The literature examines search personalisation and how intuitive it has become as it serves us with content that matches our online behaviour. We discover that over-personalisation is narrowing our search bubble which in turn can blind us from other information outside our search reach.

During the course of this study, a mixed method research was used to measure and discuss public opinion. This study has identified where literature has fallen short. Where gaps have appeared in the literature, qualitative meaning has been extracted from 3 groups of people from differing levels of technical ability and experience.

It was discovered that age demographic affects how we view Google as a search engine. Quantitative findings revealed that 75% of users were aware of personalisation but 70% of these users were not always happy with the results. Although a general consensus was formed that personalisation was helpful in many respects, it was also argued, in some camps, that it breached ethical boundaries. Overall, there was a national split whether users considered personalisation ethical or not.

This study leaves open the debate of how we perceive Google as a nation but it also gets closer to the truth of how we feel about the search engine on a personal level. The research also confirms that there is plenty of opportunity to carry out further studies into this area.

Chapter 1: Introduction

1.1 Introduction to Study

At Stanford University, in 1995, Larry Page and Sergey Brin met for the first time. Within a year of their meeting, they were collaborating on a search engine called Backrub. In 1997, their collaboration became Google. Seventeen years after their first meeting, Page and Brin had developed Google into the largest, most popular, most trusted, and sophisticated search engine in the world. With an estimated revenue of $65.8 billion in 2014 (Market watch, 2015), Google has become the leading brand associated with the Internet.

This research is an examination of the level of knowledge and trust that the Irish public have in the search giant, Google. The author examines the new semantic search engine, search personalisation, trust, reputation, user reliance on the Internet, and the ethical implications associated with this. The study also examines the monopoly Google has on user search, giving them a strong ability to influence user decision-making online.

Although having a monopoly is not illegal by any means, this opinion is illustrated in an article by Ingram (2011), where he refers to a statement by Judge Hand (1945) that “The successful competitor, having been urged to compete, must not be turned on when he wins”. According to Ingram, what can be considered illegal is obtaining that monopoly by unjust means. Ingram also claims that the problem with a company like Google, even if we were to assume it has acquired its monopoly unfairly, is the difficulty in proving that it has a negative effect on its many users globally.

In order for the author to research this topic at a deeper level, a prior knowledge of the evolution of Google search over 17 years was essential in gaining an understanding. Google semantic search has changed the way we discover information online. Traditionally, we have searched for keywords and the results Google served the user with were based on matching those specific words using algorithms.

These results were often unsuccessful and the user’s intentions were not always fulfilled. The literature review chapter details an overview of the existing literature; the methods of search Google have developed to produce more relevant results and therefore gaining trust as the number one search engine in Ireland and globally.

The literature study reveals the evolution and emergence of Google as a search engine of authority over seventeen years and the implications it has had and the role it plays in the everyday lives of the Irish public.

On examination of the literature, it is apparent to the author that there is a lack of existing studies and research in Ireland on the topic of Google semantic search personalization. In Ireland, Google dominate searches, with 95% (MVF Global 2015) of the online population choosing it as their search engine of choice. The author also examines literature revealing that there is an over reliance on Google as a Knowledge database in certain age demographics, therefore altering traditional learning methods.

1.2 Focus of Research

Google's growth in Ireland has been exponential. Ireland’s Internet population is just over 3 million, with 95% of search market share (MVP Global, 2015). This growth and market share, along with how Google has gained a high level of trust and reputation amongst users, are some of the motivations behind the author’s research into this chosen topic. The emergence of semantic search personalisation and Google's understanding of search context has been of great interest to the author, especially with the evolution of Google's search algorithm mining data for the intuitive knowledge and answer engine.

The author’s interest in the algorithmic changes within the Google search engine stems from his own experience of website ranking fluctuations due to changes in algorithms. Understanding how users view Google from a reputational and trust point of view, and investigating user opinions on Google's power of influence and ethics, are key for this research topic here.

Findings in relevant research and literature in Ireland have presented the author with an opportunity to carry out primary and secondary research into the level of knowledge, understanding, and trust that exists amongst the Irish public in Google search personalisation.

The focus of this research is to establish, through the use of surveys, interviews, and questionnaires, the differing opinions and findings between three participating groups – firstly, the general public, secondly, a group who rely on Google in their own professions, and, thirdly, a group of experts who are very familiar with Google Search as a marketing tool. The study will focus on the Irish public's knowledge of Google search, particularly at a psychological level, investigating how it makes them feel from a trust and ethical perspective.

1.3 Structure of Thesis

Chapter 1

This chapter introduces the research topic. The author outlined an overview of the literature reviewed, the research objectives and methodology, and the focus of this research.

Chapter 2

In this chapter, the author will discuss the available literature on the topic and present an evolving journey from that of Boolean search to one of semantic search. During this evolution, the literature reveals Google's transformation from a search engine to a personalised knowledge based trust engine. This chapter also identifies gaps in existing literature.

Chapter 3

The research methodology is introduced in this chapter. The author will discuss the approach that was taken, the research design, and the method in which all data was collected. An overview of the theoretical perspective taken in this study and an outline of the epistemology are also discussed. The author introduces three data collecting methods, including, an online survey, interviews, and a questionnaire. Each of the three research samples are discussed and the rationale for choosing them examined.

Chapter 4

In this chapter, the author presents the findings of the quantitative research methodology. The empirical objectives of the research are reached using an online survey that addresses all objectives of this study. The results are broken into three sections and sub sections where the raw data is presented in textual, table, and chart formats. Results are also discussed and analysed in a clear and concise method so that conclusions can be formed.

Chapter 5

In this chapter, the author presents the findings of the qualitative research methodology. The findings will be presented from two areas, technical group interviews, and an expert group questionnaire. Analysis and discussion will be made, along with a comparison between data sets. During discussion and analysis, the author will make reference to the literature review, with a view to establishing, comparing, and reinforcing opinions.

Chapter 6

In this final chapter, the author will draw a conclusion to the overall research. Reference will be made to chapters 2 to 5 in order to provide clarity on how the objectives of this study were met. Recommendations for additional research in the future will also be discussed.

Chapter 2: Literature Review

2.1 Introduction

The following chapter is a review of existing literature in relation to semantic search personalisation and the implications it has on users of the web as a source of knowledge. Online sources including conversations in specialist websites or communities (textual and video), reports, research and published books, and articles and papers make up this review. As the chosen topic is such a new area of technology within the digital marketing space, much of the literature findings were sourced online.

In order to research the above topic, the author needed to uncover a path of evolution from early Google search results, based on the combination of keywords that often times resulted in misinformation, to the now more accurate Semantic Web and Personalisation, based on understanding user search intent. Personalisation is the result of an intelligent semantically intuitive group of algorithms that link people with 'things'.

This literature review will map out the journey of semantic search, personalisation, and the ethical implications of filter bubbles and the possibility of a biased influence from within the Google platform. In the author’s opinion, this literature review can only bring this research so far. It acts as a foundation. The findings and analysis from mixed method research are the building blocks that complete this research.

2.2 The Evolution of Google Search keywords

It is important to understand the evolution of Google search from traditional Boolean search to one of semantic search. Sieber (2013) gives us an in-depth look at the principles of how Boolean Search logic works, using words such as 'And', 'Or' and 'Not', alongside search queries to refine or filter the results for users. They form a relationship between sets. David Amerland, a highly regarded specialist in the semantic search area, has written extensively on semantic search. In his book, titled Google Semantic Search, Amerland (2013) talks about the new intelligence on the other side of the screen. The word 'Semantic' derives from the Greek word for 'meaning'. This is exactly what Google tries to mine for, using your search query to excavate relevancy from the indexed web. Amerland explains that Google’s goal is to serve the user, with the most relevant search result, at the right time, considering their location, time of day, previous online activity, and also including the user’s social signals. There are many other criteria that Google base its search results page (SERPS) on.

Google's search algorithms try to understand the intent of the user and the context in which they are searching. Instead of matching keywords, the search algorithm understands long tail sentences. Amerland explains that the reason for the slow development of semantic search technology has been due to two factors, data and scalability. In order to understand user’s intent in a semantic search environment, Google's algorithms need to pull from vast amounts of data.

We have never had access to so much data as we have today. Big Data needs to be indexed in a meaningful way so that humans can understand it. Amerland (2013) explains that, for Google's semantic search algorithms to index trillions of web pages efficiently and serve users with search results within fractions of seconds, it must be able to scale across the web without human intervention.

Machine intervention is needed to index and mine the semantic web, as we know it today. In an online article, Bradley (2013) describes the semantic web as a search environment that, not only provides search queries with answers, but also provides us with connections between things.

Figure 1 - The Evolution of Google Search

illustration not visible in this excerpt

Source Mashable.com

2.3 Google Personalisation

As semantic search is now more intuitive than ever, the emphasis on personalisation is becoming more prevalent. With traditional Boolean search, we were subjected to search results that were almost identical for everyone within their own immediate location. Semantic search has made search more relevant to the user through personalisation. Early advancements in personalisation, as reported on Google Blog (2007), states that Google's goal is to give users an experience that is relevant to them personally. In the same web article, Google also mention that personalisation will initially be subtle to the user’s view but that over a period of time the search engine will learn the preferences of that same user intuitively. Over time, the search results will be become more 'filtered', thus being more relevant to the user as the engine learns.

2.4 Echo Chamber Effect

In an online article on Sciblogs, Jacobs (2011) talks about personalisation and the negative effect it could be having on user’s search results. Jacobs compares Google personalisation to the 'Echo Chamber Effect'; “Where a person or group who repeatedly only hears their own views echoed back to them [in this case through search results]”. Jacob’s concerns are with Google suggesting similar results each time, based on the user’s preferences and activity online. Jacobs believes that semantic personalisation is closing in and narrowing our search result choices, limiting exposure to other views outside of our own.

In a post on Moz blog Shepard, Cyrus (2011) reports that every personalised search result on Google returns a different outcome for everyone, depending on location, relevance, diversity, past search history, social activity, and more. There is no longer a page one. For search marketers, this can be a problem when relaying ranking positions with clients. In theory, there are two sets of page rankings, one where the user is logged into Google, and the other when logged out. Shepard offers various solutions to achieving un-personalised search results.

In his book, The Filter Bubble, Pariser (2011) remarks that, in an online world, it is becoming more difficult to escape one's own point of view. Pariser talks about the personalisation filters that Google use to shape our search results. These same filters will eventually bias the content you see. As the semantic web becomes more personalised, the search results we see become more refined until, at some stage, we find it difficult to break out of what Pariser calls the 'filter bubble'. This is where we must ask the question: is Google search personalisation of benefit to us or to Google itself?

2.5 The Filter Bubble

The Filter or search bubble is a phenomenon that has been widely spoken about. In an article on forbes.com, Amerland (2013) discusses how semantic search delivers more relevant and richer search results than any other form of search. Amerland also agrees with Pariser (2011) that personalisation is narrowing our search bubble and, in turn, can blind us from other information outside our search reach that is not served to us. According to Amerland, the ' surprise ' or the ' by chance ' element of search results we sometimes encounter may reduce dramatically. As personalisation has become more accurate and narrow, serendipitous information has been introduced to broaden the search result for the user. This can open up the user’s filtered search bubble. Amerland talks about 'programmed' or 'forced' serendipity, which is a contradiction to its real meaning 'pleasant surprise'. Google's serendipity is likened to Amazon's product suggestions; 'Customers who bought X also bought Y'.

After Parisers Ted Talk in 2011, regarding the Filter Bubble, Schwartz (2011), of Search Engine Round Table, investigated user’s thoughts on filter bubbles and asked the question of whether Google personalisation was a form of censorship? In his survey of 106 people, 77% agreed that personalisation was a form of censorship.

Pariser increased awareness of Google's personalisation, which fuelled an already hot debate on ethics and privacy. Hacker News asked Google's head of web spam, Matt Cutts, about the notion of over-personalisation. Cutts (2011) responded that, if a user wishes to see search results without personalisation, it can be turned off by adding "&pws=0 to the end of a search query or using the incognito function in Google Chrome. He also pointed out that personalisation has less impact than localisation. Most users wouldn't be unaware of turning off personalisation for less biased results. Cutts added that Google’s semantic algorithms are programmed to serve variety in search results, which relates to what Amerland refers to as forced or programmed serendipity.

Maccatrozzo (2012) claims that letting more into the search bubble and allowing it to grow cannot rectify the problem of over-personalisation. One of the problems Maccatrozzo sees with relaxing the filter bubble is that it will serve less relevant search results to the user, which, in turn, will give users a negative online experience. In 2010, during the Deepwater Horizon oil spill disaster (Wall Street Journal, 2010), Eli Pariser asked two friends to carry out a simple test (Pariser, 2011). He asked them both to search for the term “BP”. They were each served very different results from each other. One was shown results regarding investment information about BP and the other was shown links about the oil spill. This over-personalisation could in theory influence how users saw and thought about the news. Pariser maintains that, if Google understands its user’s online habits, activities, and opinions, then they are in a better position to influence that same user's decisions.

Figure 2 - Pariser's Filter Bubble

illustration not visible in this excerpt

Source - http://myredhotphilosophy.blogspot.ie/2011/05/eli-pariser-beware-online-filter.html

2.6 The Knowledge Graph - things, not strings

A post on the Google Blog (2012) introduced us to the Knowledge Graph and the concept of things, not strings. This concept describes traditional search, which was based on matching the user’s keywords (strings) with the best matching results available. The transition to semantic search provides a richer meaningful experience for the user, understanding real world entities or things (people, places, objects etc.) and how they are connected. It tries to take out the ambiguity from search queries.

Amerland (2013) also describes Google's knowledge graph as the brain behind semantic search. The search results of the past served us with best guess results for us to trawl through. With the knowledge graph integrated seamlessly into semantic search, Google has now become a knowledge engine that serves us with richer meaningful results, recommendations, and predictions. According to Google (2012), the Knowledge graph helps a user explore collections of linked data, which, in turn, enables us to research topics faster and at a deeper level.

The knowledge graph was appearing more in search results, according to Meyers (2013), who noted that, on the 19th July 2013, the Knowledge Graph served 26.7% of search results, a climb from a previous average of 17.8%. Meyers suggests that Google had lowered their algorithm threshold so that more queries would launch Knowledge Graph results. Purtell (2013) states that Google are now able to serve the users with a better experience. Users can find what they are looking for without having to read through each search result that may have alternative meanings.

Juel Vang (2013) in her conclusion, in the paper, Ethics of Google's Knowledge Graph: some considerations, remarks that, although the Knowledge Graph can be a key information service in user’s lives, it can also make the rest of the web appear superfluous. Google can conveniently serve knowledge Graph results on its own web page, keeping users from clicking on external search results. Juel Vang also maintains that the Knowledge Graph raises two concerning issues:

1. If a user is served with an immediate result to their search query, is there any need to look at other results on external pages? This surely grows Google's dominance, keeping traffic from other pages.
2. As Google has made data retrieval much easier for its users, will this affect our information retrieval skills, as we now rarely venture past the initial search result on page one.

Juel Vang adds that we are becoming a lot less dependent on using our minds to hold knowledge. We are now using technology to dump and retrieve information at will. Juel Vang questions Google's dominance as a search engine; “Should we, however, trust Google as a de facto editor of our knowledge?” (Juel Vang 2013).

2.7 Search Psychology

The psychology of search is very much an emotional process. Our decisions are made within an instant of seeing search results. Influence from within those search results can be quite strong, especially from an authoritative search engine like Google. In a research paper on the Google Generation (those of us born after 1993), Rowlands et al. (2008) noted that, although this young generation is extremely competent with technology, their information literacy has not improved. It was also noted that search results were merely scanned and not reviewed or evaluated with relevancy and authority in mind. Rowlands et al. also remarked that the Google generation prefer to search using natural language instead of analysing the keywords that may perform better.

Semantic search is more natural to the Google generation than the traditional keyword weighted Boolean search. Rowlands et al. also state that the Google Generation fail to see the Internet as a combination of networked information supplied through different providers. This failing leads to the user seeing Google or the like as the main brand associated with the Internet. This works to Google’s advantage, instilling trust based on the monopoly it has on search.

2.8 Reliance on the Internet as an external memory bank

An article by Davis (2011) cites Sparrow et al. (2011) that the Internet has become an external memory where information can be extracted at will. In a study carried out by Sparrow, 60 Harvard students were asked to type 40 pieces of trivial information into a computer. They were then told the information would either be saved or erased. Those students who thought the information would be saved to computer were less likely to recall it. The study suggests that, when people think about technology and have to recall information at a future date, they have a lower rate of remembering that data.

The study also found that people would go to the Internet as a primary source for knowledge. According to Sparrow, we easily remember where we can get the information but find it more difficult in remembering the actual information. Sparrow concluded that human memory is adapting constantly to new technology. According to an article by Latham (2011), we call this reliance on the Internet for data retrieval 'The Google Effect' or ‘Google Amnesia’. According to Latham “the brain is very plastic and adaptive and I wouldn’t say we are dumber because of the Internet. One might say that we are using the Internet in ingenious ways”.

Latham also explains that because our activity on computers and smartphones has increased, information is available to us whenever we want, lessening our need to learn information in traditional ways and heightening our reliance and trust in search engines such as Google.

Carter (2012) discusses search at its root. The reason a user turns to Google is because they have a question that needs immediate attention. Each of these search questions affects the user’s life in some way and they all share the common denominator of missing information. Carter refers to a preliminary stage of search as the Pre Population Stage. This is the stage before a search query is entered into the search box. It is the stage before the query is textualised by the user. As we are visual and conceptual thinkers, it is difficult for us to translate this stage into keywords for search engines. There is a disconnect between these two stages but, as semantic search understands our intent and context to a certain degree, it proves Google is now paying attention to the way we think. Search is psychology and the Semantic web is learning from us all the time. The moment we search online, we are at our most susceptible to influence or suggestion.

2.9 User sentiment

According to Monson (2008), in an interview with the Google Search User Experience Team Leader, Jim Wiley, search is very much part of psychology. Google understand this. Wiley explains that a user’s feeling or sentiment towards a service, brand, or product stems from their own self esteem. When searchers find the User Interface of a website difficult to navigate, this can cause frustration and the user can feel bad about themselves. On the flip side of this, Wiley points out "When a UI (user interface) is intuitive enough for beginners to quickly become power users, they feel empowered and are more likely to invest themselves in that product or service". The author reflects that, if Google can understand and gauge the sentiment of user search based on intent, context, previous search history, and many other factors, they can then apply all of these to an algorithm that could have psychology as its foundation.

2.10 Search Influence

With psychology entrenched in Google's strategy, how much influence does the search engine have over our opinions? In a recent report, Prigg (2014) asked the question of whether Google could fix an election. According to Prigg, researchers at the American Institute for Behavioural Research and Technology analysed the results of an Indian election. They found that a high proportion of undecided voters cast strong attention towards higher ranked search results. The research has shown that voters can sway opinion in favour of one candidate or another based on search rankings.

One of the researchers, Dr Robert Epstein, claims that, if two candidates were battling for higher search rankings, it would be fair. He suggested that, if Google were to be biased towards one candidate, they could be successfully elected based on the outcome of Epstein's research. This sway in opinion from undecided voters can be as high as 12% and, in a country with a vast population like India, could lead to a win margin of up to 3%. The researchers say that 25% of national elections throughout the globe have winning margins of less than 3%.

The study revealed that certain groups within the demographic were more easily influenced than others. Epstein also discovered that 99% of the people involved in the study didn't realise the search results were biased. This suggests that Google could manipulate or influence people's opinions without them even knowing. Could Google be guilty of serving users with search results irrespective of responsibility and ethics?1

Another influential channel is remarketing within the Google Adwords platform. Users can be targeted based on their interests and the websites they have visited. Whenever a user comes to a website set up with a remarketing code, an anonymous browser cookie is dropped. Later when the user revisits the web, the cookie communicates with the retargeting provider and ads are then served. According to an article by Couch (2013), remarketing enables marketers to take advantage of users, as they've already visited a website and have shown interest and considered an action. Couch also explains that remarketing ads can have a higher conversion rate than 'first seen' ads, as they can act as a reminder or persuader to convert a user to buyer. Couch does warn that excessive remarketing can be damaging, especially from high profile brands. Users can find these annoying, therefore 'devaluing' the brand. According to Couch, although remarketing can be successful within an overall digital strategy, the conversion rate diminishes after 8 viewings of an ad by a user.

When it comes to influence over online behaviour, we can take a look at the many products that Google have launched. Search is by far the most integrated and successful. Over 90% of their revenue stems from search. Search is marketing and less intrusive than traditional methods. Search allows users to seek out products when they want them. With this, Google has the power of influence over its users. Semantic search is possibly the most radical change we have seen on the web in the last 15 years. If personalisation can lead to limited exposure to information outside of our own opinion, then what are the implications? Semantic search allows us, as users, to open up to Google in a way we have never before. With this openness comes vulnerability and with this vulnerability comes susceptibility to influence.

2.11 Ethics

In all aspects of semantic search, we must consider ethics and responsibility. If Google is to influence our choices, then let it be on a fair and level playing field. Hinman (2008) points out that search engines are opinion influencers, as they “contribute significantly to the social construction of knowledge”. Hinman also added that, not only do search engines like Google provide gateways to knowledge, but they also “play a crucial role in the constitution of knowledge itself”.

Tavani (2012) asks if companies such as Google should have heightened moral obligations because of their “privileged place” in a society that depend on the Internet, both socially and economically. Many authors who have written about the ethical responsibilities of Google and other search engines, referring to them as “Gatekeepers of the Web,” and this would suggest that much responsibility should be expected of them.

With content added to the internet every day at an exponential rate, what epistemic value can Google put on the web? Simpson (2012) describes the epistemic role that search engines play in data retrieval as likened to that of a 'Surrogate Expert'. Without help, a user searching for information is unlikely to find what they are looking for quickly.

Goldman (2008) questions the epistemic value of search engines such as Google, stating that anyone can become an author on the internet, with their data readily available to a vast audience. To Goldman, this is highly unprecedented. Goldman also questions the veracity of web content and whether we know if it is epistemically superior to traditional offline media, which would normally have to meet stringent publishing criteria. Although Google can be seen as the primary influencer of search, the results it provides us with must follow an ethical code of responsibility that has a user’s best interest in mind.

Hamilton (2012) asks if Google can conduct business in an ethical fashion, if it allows users to research unethical or unlawful content on its platform. If Google is the Gatekeeper of Knowledge how much data should it censor? What is clear is that new European legislation now protects the user and recognises ownership of online information and identity.

In an article on Silicon Republic, Toby (2015) reports that Google has received over 281,000 requests to remove over one million links from its index. Google removed over 600,000 links in order to comply with the Europe-wide ‘right to be forgotten’ legislation. The problem with this legislation at the moment is that, although requests were removed from google.ie, google.co.uk, google.fr and other European Google domains, the links are still traceable on google.com.

2.12 Google's Monopoly

Google’s monopoly in the digital space can be likened to a major oil company controlling the production of oil. This could impact the global economy. Google’s mission statement is to “Organise the Worlds information”. According to NetMarketShare (2014), as of December 2014, Google had control of 66% of global market share and 95% of Irish market share (MVF Global, 2015). Users need to demand that this data organisation is carried out in an ethical and transparent fashion, protecting user’s rights to privacy.

2.13 Antitrust

In April 2015, Google were accused of distorting its search results in favour of its own shopping service. In an article in Engineering & Technology Magazine (2015), the EU’s Competition Commissioner, Margrethe Vestager, stated that:

I am concerned that the company has given an unfair advantage to its own comparison shopping service, in breach of EU antitrust rules. If the investigation confirmed our concerns, Google would have to face the legal consequences and change the way it does business in Europe.

This investigation started back in 2010 and, if Google are deemed to be found in breach of EU competition laws, they could be faced with a company 'break up' and fined up to 10% of revenue (€5 - €6 billion euro). As search engines record our time online as digital signals and build a personalized search experience, they must serve us with the results that best suit our requirements. Open mindedness and serendipity are both factors of being human. Should Google embrace these ideals and serve us in a fair an ethical way through their search algorithm?

In a recent article on Digital Monopolies, Ginger (2014) takes a pro Google stance on their dominance as a search engine. Ginger applauds Google's business model and asks why innovation that is clearly at a higher level than its competitors should be penalised. Ginger further adds that, not only is competition fair, it should not be regulated in such a way that deems it unfair. Ginger concludes his article with:

What is important is to ensure that these companies do not use their large databanks to exploit other industries where they can leverage their power in unfair manners and create unnatural monopolies.

2.14 Trust &Reputation

Pan et al. (2007), in their research paper, In Google We Trust, cite Luhman (1989):

"Trust is a mechanism humans use to reduce the complexity of decision making in uncertain situations". According to Gigerenzer & Selten (2002), in reference to search results, the user may feel they are discovering the information themselves, therefore promoting the idea that Google's influence on trust can work also on a subconscious level. This could be viewed as exploitation when taking into account the regularity of the Information environment.

Pan et al . carried out an eye tracking lab study based on a subject group of 22 students. The students were asked to locate predetermined websites and specific information using only Google as a means to search. The objective of the study was to identify if users clicked on search results based on the position they appeared on the search engine results page (SERPS) rather than result relevance. Unbeknown to the students, the authors randomised the search results using a proxy server running Google search. Google ads were omitted from results with only organic results presenting.

The experiment provided search results in random order not necessarily showing the most relevant results in order 1-10. The findings showed that students were heavily influenced by the order the search results were presented to them, irrespective of how relevant they were to the search query. Pan et al. noted that trust in Google search was very high amongst this sample audience and, even though there may have been some conflict in result position over the searcher’s own evaluation, it still wasn't enough to override Google's influence and user trust on result clicks.

According to Grumen (2014), Google is the second most valuable brand globally. Grumen adds that Google's business model is to mine data to serve its users. He also points out that “Google Search has always been an ad-supported service, so it needs a way to sell those users to advertisers". As many of Google's trusted products are free of charge, Grumen points out that we pay with our personal information and that nothing is free. He argues that Google's vision in the beginning was user centric but now that it has shareholders to satisfy, a tension between reputation and 'economic reality' has been created. Grumen claims that if Google ignore this tension, user trust in the brand could diminish.

The extent of Google’s dominance online is built on a foundation of user trust. In a recent article by the Irish Times (2015), it was reported that a survey conducted by the Annual Ireland RepTrak® 2015 study and compiled by the Reputations Agency (2015) named Google as the most reputable and trusted brand in Ireland. The survey was conducted using a sample of 4900 members of the Irish general public. According to RepTrak, “Rankings measure the public’s perception of companies based on four emotional indicators - trust, esteem, admiration and good feeling”. In the same article, the managing director of the Reputations Agency, Niamh Boyle explains that:

Reputation matters for companies - with strong reputations helping to win custom, attract the right people, gain support from key stakeholders and ultimately drive business performance.

The head of Google’s Irish operation, Ronan Harris, reinforced Boyle's statement on reputation by adding: “A good reputation is essential to any company, but it is fundamental to a company like Google. In today’s world, consumers will only transact with you if they trust you”. This statement from Harris relays the importance of trust to Google and their online model.

2.15 Conclusion

This literature review has outlined and documented the evolution of Google search from its beginnings to a now more sophisticated semantic search model and, according to Amerland (2013), it is driven by a new intelligence behind the screen. The author has outlined a pattern of over reliance on Google as an information dump. This has changed traditional learning methods as we knew them to a new digital reliance that provides users with instant answers, making memory recall unnecessary in many cases. Millennials see the internet as a single brand according to Rowlands et al. (2008) and this, in turn, benefits Google; it is now the search engine of choice for 95% of users in Ireland (MVF Global, 2015). With this monopoly in the search market, users place great trust in Google as a gatekeeper of knowledge.

Ethical responsibility is as important as the information Google serve its users. There is a gap in the literature from an Irish perspective, as much of the previous research is presented from outside of Ireland. This gap in literature identifies opportunities to carry out further study in this topic. Further research into the privacy and antitrust areas would be most welcome as, at the time of writing this thesis, Google were under scrutiny from the EU for a possible breach of European antitrust rules and an outcome had not yet been reached. With Google’s dominance of search marketing in Ireland unquestionable, and the level of trust they already command in search, the author will provide, in the chapters that follow, quantitative and qualitative research findings to fill in the gaps that exist in the literature from an Irish study perspective.

Chapter 3: Research Methodology

3.1 Introduction

This chapter focuses on the approach taken by the author within the research design and methodology. A mixed method approach has been used throughout this research. The various tools and methods used in this research are highlighted throughout the chapter. In this chapter, the author addresses the dissertation topic by guiding the research accordingly, resulting in an outcome that is both satisfactory and unbiased, and one that is a true reflection of the quantitative and qualitative methodologies used. According to Crotty (1998), there are four questions that are fundamental to any research methodology:

- What methods do we propose to use?
- What methodology governs our choice and use of methods?
- What theoretical perspective lies behind the methodology in question?
- What epistemology informs this theoretical perspective?

Crotty's four questions can be approached in reverse order, as they are integral to the research and form a process where each question will have an informative relationship with one another - Epistemology, Theoretical Perspectives, Methodology, and Methods.

Figure 3 - Relationship between epistemology-theoretical perspective-methodology-methods

illustration not visible in this excerpt

Source: Crotty (1998)

3.2 Epistemology

De Rose (2005) defines Epistemology as the “theory of knowledge”. He elaborates further, stating: "Epistemology, then, is the branch of philosophy that deals with questions concerning the nature, scope, and sources of knowledge". In one of Crotty's definitions, he simply describes epistemology as "How we know what we know". Gray (2009) states that “Epistemology provides a philosophical background for deciding what kinds of knowledge are legitimate and adequate”. Gray also remarks that Epistemology is to understand "what it means to know".

Epistemology can be separated into two perspectives: positivism and relativism. According to Easterby-Smith et al. (2003), positivism and relativism work with one another when the researcher is tasked with identifying the “pre-existing reality”. Easterby-Smith et al. state that the positivist researchers approach is "understanding reality through design of experiments", but the relativist achieves it through the use of a “triangulation of methods", meaning the use of two or more research methods and applications to study the same phenomenon. Triangulation forms the foundation of this research methodology. It adds revealing layers and depth to the knowledge at an experiential level.

Crotty informs us of the three positions or views within epistemology: Objectivism, Subjectivism, and Constructivism.

1. Objectivism

Crotty refers to objectivism as the existence of knowledge whether it is known to us or not. He also adds that this view is the framework, in a traditional sense, for theoretical perspectives. Researchers with an objectivist view look for causes and effects and, at the same time, search for explanation in a phenomenon.

2. Subjectivism

The epistemology of subjectivism, according to Crotty (1998), is that the meaning of anything comes from a vacuum. Gray (2009) agrees with Crotty's interpretation of the relationship between object and subject when he explains that meaning is not a factor that emerges from interaction between subject and object; in fact, the subject imposes itself on the object and therefore suggests that the object on its own plays no part in contributing to meaning.

3. Constructivism

Constructivism, according to Crotty (1998), is the idea that "truth or meaning comes into existence in and out of our engagement with realities in our world. Meaning is not discovered but constructed". The true sense of the knowledge is discovered after engagement with people and observations found in analysing online surveys, strategic interview conversations and online interviews. An overall understanding will be constructed from three research methods, including interaction and observations of people during conversational interviews, online surveys, and a questionnaire with industry experts.

3.3 Theoretical Perspective

The theoretical perspective, according to Crotty's research design framework, is defined "as the philosophical stance informing the methodology". Crotty claims that there can be several theoretical research views resulting from differing epistemological stances.

Gray (2009, p. 14) questions whether “in research should we begin with theory, or should theory itself result from the research?” Gray acknowledges that, in testing the explanation, an idea or phenomenon can be verified, disproved, or reformed. The author suggests, in the case of the objectives of this dissertation, ' that theory should in fact come from the research'. The interpretation of the resulting data from this research will form theories that will address many of the objectives set out by the author.

Crotty (2003, p. 5) states that a research design can have more than one perspective such as positivism and interpretivism. Jacobson (2013) explains that “Positivism in general refers to philosophical positions that emphasize empirical data and scientific methods". In opposition to positivism, is interpretivism, and Abbot (2012) argues that, to understand social action from an interpretivist perspective, scientific methodology will not work and a qualitative stance is needed.

3.4 Research Methodology

Rajasekar et al. (2004) state that research methodology is the use of a systematic structure when solving a problem or addressing a phenomenon. In other words, it is the scientific process we use to carry out research. Rajasekar et al. also defined methodology in simple terms as "the study of methods by which knowledge is gained".

This dissertation uses phenomenology in its approach. According to Boyd (2015), phenomenology "involves trying to understand the essence of a phenomenon by examining the views of people who have experienced that phenomenon". Phenomenology is interested in the experiences of individuals and can involve in-depth interviews as this dissertation will show in its method. The author uses the experiences found in interviews, expert conversations, and survey commentary to provide a well formed phenomenology.

Figure 4 - Conceptual Framework

illustration not visible in this excerpt

Source: Ivory, P. (2015)

3.5 Research Methods

There are many research method procedures that can be used in study. The use of surveys, interviews, observations, questionnaires, and sampling are very common research methods. According to Rajasekar et al., choosing the appropriate research methods are essential when collecting data and finding a solution to a pre-defined problem.

Crotty (2003, p.5) offers many more samples of research methods, for instance, data deduction, case studies, and focus groups. The methods used by the author in this particular research include online surveys, strategic conversational Interviews, an expert questionnaire, and qualitative observations. These methods, although diverse, are an essential combination that will draw the most valuable data from different sample groups in order to produce a satisfactory hypothesis.

3.6 Quantitative Research

Watson (2015) describes quantitative research as an investigative insight of a social phenomena, using numerically driven or statistical data. The basis of any quantitative research is measurement and it also assumes that the researched phenomena can be measured.

Table 1 - Attributes of Quantitative Research

illustration not visible in this excerpt

Source: Ivory (2015)

3.6.1 Online Survey

The initial method of research in this dissertation is an online survey, using the Survey Monkey platform. The rationale in choosing this quantitative method of research is reinforced in the analysis by Isaac and Michael (1997):

To answer questions that have been raised, to solve problems that have been posed or observed, to assess needs and set goals, to determine whether or not specific objectives have been met, to establish baselines against which future comparisons can be made, to analyze trends across time, and generally, to describe what exists, in what amount, and in what context.

According to Kraemer (1991), the three main characteristics of survey research are:

1. The first characteristic can be used to describe the specific facets of a cross section of population on a quantitative level. These can include understanding the relationships between variables.
2. The second characteristic refers to the collection of data from participants and that it is of a subjective nature demonstrated clearly by the participant’s point of view.
3. The third characteristic of survey research uses a section of population where the results can be scaled and looked at in general terms of the entire population.

3.6.2 Data Collection - General Public Online Survey

The survey includes ten questions, which fall into four categories: Personalisation, Influence, Trust, and Ethics. Age demographic was also an important factor as comparison between different age groups provided varying opinion, as will be shown in chapter four. Two of the questions allow for additional comments, so participants could elaborate on their answers. This commentary will be used in chapter five to help analyse data findings.

Within the online survey, participation was promoted over social media channels and email. Facebook and LinkedIn were the primary channels, as the author felt shareability was more likely here from a social perspective. A web link was embedded within numerous Facebook and LinkedIn posts and published at times where highest engagement could be achieved. The author has considered the recommendations of this study and published survey related posts on Facebook during the times of 1pm and 3pm on Thursday and Friday. On LinkedIn, suggested times differed to Facebook. Tuesdays between 10am and 11am proved more likely to gain the most clicks and shares. Google plus and twitter were secondary channels where a link to the survey was shared.

Figure 5 - Best day and times based on maximum engagement to post on Facebook

illustration not visible in this excerpt

Source: Mashable.com

In a research paper, Fundamentals of Research Survey Methodology, Glasow (2005) refers to Salant and Dillman's (1994) observation, that selected participants for any survey sample should be selected at random and must have an equal chance of participation. The author is in agreement with Salant and Dillman while still adhering to a strategic and structured best time to post policy. Randomisation in the context of this survey refers to the author not targeting a specific age demographic, as it was felt that four defined age demographics would be easily reached with post and status promotion (see Figure 6 below).

Figure 6 - General Online Survey Age Demographic Participation

illustration not visible in this excerpt

The initial objective of this online survey was to reach a sample audience of 200 participants. On reflection, when 200 was reached, it was deemed too small a sample, as the author wanted to harness a wider and more weighted age demographic, as this would prove an important factor in later findings and analysis. A new objective of 300 participants was set and, when the survey was closed, 310 participants from the general Irish population spanning four age demographic groups had answered ten questions.

It was decided that, with over 300 participants, the survey was more viable and its generalizability to the broader population was better. The author agrees with Babbie's assumption that generalizability enables us to look beyond the limitations of our own research data (Babbie, 1995). According to Babbie, generalizability also makes us feel that our research has more weight or more importance.

3.6.3 Data Collection - Technical and Expert Group Surveys

The technical and expert group surveys were identical to that of the general public survey but served a very different purpose. Both surveys included 4 participants each. For the purpose of this dissertation, the author refers to these groups as 'the technical group' and 'the expert group'.

Both survey samples are too small for any generalizability. The author's rationale in surveying these groups is self serving. Comparing the tech group and the expert group with the general public group enabled a deeper understanding and interpretation when it came to analysing the findings of their input to the research.

[...]


1 A full comprehensive list of the research and results are available at http://aibrt.org/POS-real_time_display/rt_detail.php, AIBRT (2014).

Details

Pages
103
Year
2015
ISBN (eBook)
9783668208315
ISBN (Book)
9783668208322
File size
4.7 MB
Language
English
Catalog Number
v320189
Grade
First Class Honours
Tags
personalisation trust influence reliance reputation ethics google semantic search irish discussion

Author

Share

Previous

Title: Google Semantic Search in Ireland. Personalisation, Trust, Influence, Reliance, Reputation, and Ethics