Harnessing The Power Of Social Media And Web Analytics Pdf

harnessing the power of social media and web analytics pdf

File Name: harnessing the power of social media and web analytics .zip
Size: 1556Kb
Published: 18.06.2021

Search & Content Analytics

JMP Discovery Conference Relative to the global population, there are now twice as many sensors e. The resultant explosive growth in data volume represents great opportunities for those who can find a signal in this vast collection of unstructured and uncertain noise. A recent project presented the need for JMP users to efficiently search and analyze large volumes of open source material on innovative technologies, associated maturity levels, current applications and primary competitors.

Unique challenges included the need to extract the text from these diverse website architectures absent the extraneous code and from attached PDFs and PowerPoint files that are not always in a usable format.

Once the corpus was developed, the text was imported into JMP to quickly identify critical themes of interest and associated documents. These steps were accomplished using an intuitive JMP interface to text mining routines in R.

Text analytics using JMP data visualization, cluster analysis and decision tree capabilities enabled better characterization of the market over traditional methods. This talk will demonstrate a series of JMP scripts and platforms to effectively search scores of relevant websites and monitor social media outlets such as Twitter , assemble a large collection of unstructured data, and answer focused research questions. Though the term competitive intelligence CI has only been used in the last few decades to describe the acts of a corporation to acquire information about other organizations posing business threats, the practice has been around for centuries.

Analogous to military intelligence, we can think of a business focusing on the external forces of competitors, products, and customers to accurately assess what strategy it needs to execute in order to thrive in challenging environments. We can think of competitive intelligence in terms of both the strategic environment governed by economic and market forces as well as the tactical intelligence required everyday e.

Counter-competitive intelligence is where business are also going to want to protect their own proprietary and other sensitive information. Tools and data sources need to keep pace with the dynamically evolving methods to effectively gain knowledge of business rivals and anticipate their future responses. This paper illustrates the value of JMP software from SAS Institute to both acquire and analyze data in the competitive market analytics space. Competitive Intelligence Cycle:.

The competitive intelligence cycle in Figure 1 is one of many constructs based on the military intelligence model to describe the process of finding exploitable market information on other businesses and applying analytics to discover actionable information. It starts with the definition phase figuring out exactly what dimensions you really want to investigate so you can develop a solid plan of attack.

The second phases focuses on obtaining the correct data sources to shape the research. The third phase seeks to apply analytical methods to transform the raw data into insights ultimately producing a compelling story describing the landscape.

The last phase involves feeding this information to the right teams who can act upon the intelligence.

Figure 1. The Competitive Intelligence Cycle. Data is what fuels the CI process. In the past, only CI professionals had access to meaningful data and sites. Much of the raw data will come from internet searches followed by careful extraction of the pertinent data elements.

CI collection efforts will necessarily have the challenge of unstructured text data from disparate sources and formats requiring significant effort to assemble a coherent data set for analytics. The analysis and production phase typically organizes the multiple sources of data into any of a number of constructs.

A PESTLE method evaluates the political, economic, social, technological, legal, and environmental dimensions for an industry and rivals. Other analytical frameworks form matrices with rivals as the columns and key performance indicators such as distribution channels, technological edge, pricing, market share, customer focus, financial stability, workforce, facilities, partnerships and so forth as the row.

The challenge with any of these frameworks is accurately populating the data elements as many are qualitative assessments dependent how well you can find credible data sources. We discuss a few methods in JMP that can help find and explore this critical data.

Co mpetitive Intelligence Data from Website Analyses:. He breaks these down into the site-centric which look at competitor websites versus eco-system centric that looks at the industry and specific technologies.

These tools provide detailed information about each site e. On the plot are letters that choose a representative story from the web at that point in time that may provide insight as to why there is a peak or valley. Additionally, the top geographic regions and correlated search terms e.

Google Trends can export the data which allows us to use the capabilities of JMP for our own analyses. It is not uncommon to hear in sports circles that golf as a sport is declining in interest and maybe Spieth can turn this trend around.

The plot in Figure 3 certainly shows not only a downward trend, but also a very cyclical pattern that may be of use in our competitive analytics and marketing roles. Figure 2. Figure 3. From the data export, we can use JMP to create our own custom time series plot in Graph Builder that labels those events more clearly and we can forecast the growth in the future using seasonal ARIMA as shown in Figure 4. Corporations can use these types of advanced analytics to predict where the markets are going and timing of advertising campaigns.

Figure 4. Competitive Intelligence Data from Social Media:. The last decade has seen an explosion in a new kind of intelligence data that can be assimilated from social media sites.

Communications of all types are being recorded now more than ever and we have the opportunity to find CI signals amongst all this noisy, unstructured data. Blogs and bulletin boards are outstanding hunting grounds for gathering industry and company-specific information. You can follow companies, employees, technology groups, job opportunities, former employees on LinkedIn and other professional sites. Posts to sites like Facebook, Twitter, and Instagram also offer opportunities to understand trends in markets and competitors.

The challenge is efficiently collecting this unstructured data characterized by a new informal English-like language with many ways to express the same word—think of how you would text someone the word tonight.

We wrote a JMP Scripting Language JSL script that calls the statistical programming language R to download all or a random sample of the Twitter feeds for a specified duration relative to a key word.

The script collects data from the present time to a pre-specified period in the future, although Twitter archives to collect retrospective data do exist for a cost. The data returned includes the actual Tweet, location, time, and other demographic fields. Many companies collect real time Tweets to see public reaction to major product releases—whether their own or the competition.

As an example, we ran the script for 5 minutes after LeBron James announced he was returning to Cleveland, as all of our Twitter data on products and companies is proprietary. The resulting wordcloud in Figure 5 gives insight into the most frequently occurring terms largest font after removing common words the, and, or, … and offensive language.

Additionally, we conducted sentiment analysis on the Tweets by cross-tabulating a list of approximately 1, words generally associated with positive sentiment, a list of approximately 2, negative words with the all of the words collected across 4, Tweets in that 5 minute period.

A single Tweet may have multiple positive and negative words or it may have none of either. The grand tally was 3, positive words and 1, negative words. The wordcloud and the sentiment analysis are conducted using a JMP Add-In for text mining that also calls R for text analytics routines.

Figure 5. One method to get this data is to manually copy and paste the text of interest. Alternatively, an automated web scraping utility has the advantages of being much quicker and comprehensive, though too much or extraneous information may be returned. We developed JSL code to scrape the text data from any website while keeping track of metadata like the location of the text on the website. The Spider script once again calls R and only requires you to put in the website of interest.

You can choose the type of scraping you want to conduct since websites differ in construction. Scraping options include selecting all the text, a news story or the default setting. You can also select how deep you want to crawl where level 0 is the webpage you specify, 1 is that page plus all linked pages, and so forth. As an example, a popular green energy technology is liquid desiccant air conditioning which achieves a cooling effect by removing water molecules rather than the conventional chilled air or refrigerant systems that remove humidity by lowering temperatures below the dew point.

A quick Google search on commercial liquid desiccant air conditioning shows two leading companies, Advantix Systems and Kathabar. Figure 6 displays the Spider interface along with the resultant word cloud from the text mining Add-In. Figure 6. The wordcloud once again represents the frequency of the words.

Though this raw word count is useful, it does not provide much intelligence as to what words occur together and what themes may be present. To translate unstructured text into a flat file for statistical analysis beyond word counts, it is common to form a Document Term Matrix DTM that has each document webpage for our example as a row and every unique word that is used across the entire corpus as a separate column.

The matrix entries are the frequency each term occurs in the particular row though it may be better to just use a binary representation indicating if the word appears or not.

A weighted scheme based on the number of occurrences the word has across the rows is another commonly applied transformation. That is, a word that appears in every row may not have much discrimination power to be helpful analytically. The problem with the DTM is that it is usually very large and very sparse; therefore, standard statistical methods may fail to detect useful relationships.

One solution to the large and sparse DTM is to project it into a much smaller dimension space much like principal components analysis PCA. A better solution is taking the singular value decomposition SVD of the DTM in a lower dimension than the number of columns or words reduce it to say to The eigenvectors in the U matrix represent the documents, the eigenvectors in the V matrix represent the words, and the singular values on the diagonals of the D matrix are from the eigenvalues that explain the proportion of variability explained for each SVD column.

Looking at a bivariate plot of the first two eigenvectors from the V matrix is helpful to pick up major themes in the text data as well as find words that group together. Figure 7 shows the bivariate plot of the first 2 SVDs for the Kathabar example.

Here we see the words grouping around providing a custom liquid desiccant solution and another group suggesting low energy equipment to control air temperature and humidity. Figure 7. We can also use multivariate distances from the eigenvectors of the V matrix from the reduced rank SVD. These distances show words that group together which allows you to choose any word in the corpus to find closely associated terms.

In right panel of Figure 7, we can see the words closest to desiccant are not surprisingly: liquid, system, process, dry, product, application, reliable, and manufacture.

Competitive Intelligence Data from Patents :. Fortunately, there are many great patent search engines available that give you free access to all of the abstracts, drawings, and text fields to collect data on the competitors protected property and understand the state of technology. Each paragraph of an abstract formed a separate row in the JMP data table. Figure 8. Figure 9 highlights some of the word pairs that are highly correlated like media filter, strongly corrosive carry-over, microporous membranes and so forth.

Figure 9. Topic extraction using the JMP text mining add-in provides additional information. The plot of the first two SVDs in Figure 10 gives the general idea of what these patents are trying to do: a technology that requires the use of liquid desiccant stop word to dehumidify the air resulting in cooling effect that is most applicable in humid climates; the risk is most liquid desiccants have is the use spray over a filter media that increases risk of carry-over into air space; these patents use a membrane to eliminate this risk.

Please Try a Different Browser

Social media has opened several new marketing channels to assist in business visibility as well as provide real-time customer feedback. With the emergence of new internet technologies, businesses are increasingly recognizing the value of social media and web presence in the promotion of their products and services. Harnessing the Power of Social Media and Web Analytics documents high-quality research to empower businesses to derive intelligence from social media sites. This publication is ideal for academic and professional audiences interested in applications and practices of social media and web analytics in various industries. In this volume, management, marketing, and computer science specialists, as well as other researchers from North America, New Zealand, India, and Europe, contribute 11 chapters on theories, techniques, and models of social media and web analytics for businesses. They detail the various analytical applications available to social media marketers, including online marketing strategies that can be enhanced by social media and web analytics; emerging trends, opportunities, and challenges involving social media; the effects of social media participation on consumer purchase behavior; and how three Indian firms implemented social media marketing strategies. Buy Hardcover.

Web 2. Francisco J. Mata 1 and Ariella Quesada 2. The spectacular development of the Web 2. Marketing is one of them and businesses have decided to experiment with this new type of technology in support to their commercial activities. However, to take advantage of Web 2. This requires putting Web 2.


Social media has opened several new marketing channels to assist in business visibility as well as provide real-time customer feedback. With the emergence of.


Managing and Leveraging Workplace Use of Social Media

This will navigate you to Accenture. What knowledge and insights are trapped in your unstructured data? Find better answers with AI-powered search and analytics solutions. As data grows, insight-driven enterprises need to reinvent their data supply chains to stay agile and create competitive advantages.

Remember Me. Register Lost your password? In , American Express decided to leverage big data to take advantage of this huge proprietary asset to deliver innovative products in the payments and commerce space that provide value to customers. Amex follows a closed loop system as compared to Visa and Master Card wherein it issues its own cards through its banking subsidiaries, acting as both the issuer and acquirer. AmEx is thus, able to analyze trends and information on cardholder spending and build algorithms to provide customized offers to attract and retain customers and leverage this information to maintain relationships with merchants using targeted marketing to match merchants with the right customers, who are likely to spend more and stay loyal.

American Express : Using data analytics to redefine traditional banking

Solving the Data Dilemma with Intelligent Operations

Skip to search form Skip to main content You are currently offline. Some features of the site may not work correctly. DOI: Social media has opened several new marketing channels to assist in business visibility as well as provide real-time customer feedback. With the emergence of new internet technologies, businesses are increasingly recognizing the value of social media and web presence in the promotion of their products and services. Harnessing the Power of Social Media and Web Analytics documents high-quality research to empower businesses to derive intelligence from social media sites. View PDF.

Members may download one copy of our sample forms and templates for your personal use within your organization. Neither members nor non-members may reproduce such samples in any other way e. Scope — This article provides an overview of the use of social media by employers and their employees.

Текст, набранный крупным шрифтом, точно на афише, зловеще взывал прямо над его головой: ТЕПЕРЬ ВАС МОЖЕТ СПАСТИ ТОЛЬКО ПРАВДА ВВЕДИТЕ КЛЮЧ_____ Словно в кошмарном сне Сьюзан шла вслед за Фонтейном к подиуму. Весь мир для нее превратился в одно смутное, медленно перемещающееся пятно. Увидев их, Джабба сразу превратился в разъяренного быка: - Я не зря создал систему фильтров. - Сквозь строй приказал долго жить, - безучастно произнес Фонтейн.

2 COMMENTS

Quincy F.

REPLY

JMP Discovery Conference

Kichramorbo

REPLY

Request PDF | Harnessing the Power of Social Media and Web Analytics | Social media has opened several new marketing channels to assist.

LEAVE A COMMENT