A First Tuesday Profile

Welcome to the November 2015 “First Tuesday” feature.  On the first Tuesday of every month we feature a one-on-one discussion with a person behind one of the previous month’s Best of the Business Web e-letter selections whom we felt would be particularly interesting to talk to.

This month we have chosen Tara Calishain, the creator and editor of ResearchBuzz, one of our favorite blogs for doing better online research. We chose ResearchBuzz as one of our September 2015 Best of the Business Web selections, describing it as “lively, fun and extremely informative.”

The following is an edited summary of our email-based discussion with Tara.

Q. Tell me a bit about your background--are you an MLS? Have you worked in libraries?  
 
A. I am not an MLS. I have no college diploma. I am a high school dropout; the only letters you will see after my name are GED. 
 
However I have a great respect for what it takes to get an MLS so I make it clear whenever anyone asks that I am not a qualified librarian. My mother got her AAS in library science initially, and then in the 90s got her BS and then MLS. I saw the work she had to put into it and it's no joke! Librarians, be proud of your MLS. 
 
My mother was a librarian and my grandmother worked in a book bindery, so books are in my genetics somewhere, but nowhere in my long, strange career path have I worked in a library. I did start on the Internet very early, and equally early became fascinated with the problems of organizing and finding things online. 
 
 
Q. What led you to begin ResearchBuzz back in 1998? What problem were you hoping it would solve?
 
A. I started ResearchBuzz after I wrote the 2nd edition of Official Netscape Guide to Internet Research. I wasn't trying to solve a problem so much as continue the process of keeping up and covering search resources that I find interesting. 
 
Q. Who do you think reads ResearchBuzz?  
 
There are a lot of librarians, journalists, and teachers. But it's all over the map because the resources I mention are all over the map. I don't care what you do for a living, if ResearchBuzz helps you out, that's awesome!  
 
Q. How, if at all, has the primary mission/focus of ResearchBuzz changed over the years? 
 
A. Early on I did a lot of coverage of how to use search engines better -- tricks and hacks you could use to make the most of advanced searches. (I mean hacks as in "nifty things", not hacks as in "bad things".)  
 
I still do some of that but not as much. I make more of an effort to find and cover more university/ academic/government resources that might not get as much love. For example, I recently found out about a new database of languages in New Guinea. Now, that's amazing. New Guinea is one of the most linguistically-diverse places in the world, and a database of its languages is great news. But it's not one of those things that's going to get picked up by Reuters or covered in Mashable. That doesn't mean it's not important or useful. 
 
There's a lot of wisdom online that says "do viral stuff" -- do stories which will get popular and lots of clicks and go all over the Internet. And if that's what you want to do, fine. But I want to link to things which inform you and occasionally make you laugh and HELP YOU. I want to put up a post which makes people go, "Oh man, I could use that for my project!" or "Wow, Sally could use that for her computer program!" or "Bill needs this for the medical research he's doing." 
 
My dream job is this: being able to connect every resource I find with someone who can use it. That's all. There is so much going on out there; all kinds of people are out there busting their asses to build all kinds of things. I want them to get recognition, and I want their work to be used. 
 
Q. Is there a certain topic related to effective Web research that you feel you need to revisit often because it is so important or commonly misunderstood?
 
A. Three things, if you'll indulge me: 
 
1) THE INTERNET DOESN'T HAVE EVERYTHING: It just seems like it does. There are huge numbers of items which haven't been digitized, haven't been organized -- haven't even been found! Did you read about the library which, just this spring, found a trove of ancient Greek coins in its archives?  There is still so much to do and so much to organize. It's early days yet. 
 
2) THE INTERNET IS INCREDIBLY FRAGILE: A publication goes out of business and the archives vanish. Finito. Someone decides they don't want to do their blog anymore and so they stop paying their host, or they just take it down. Those archives are gone. 
 
It's so heartening to hear more discussion about digital impermanence, to hear that the Wayback Machine is getting an upgrade, or to read about a library teaming up with a literary journal to house its archives. The problem of the impermanent Internet is only recently the subject of widespread discussion (librarians and archivists of course have been aware of it for years!) and it's a long way from being solved. 
 
3) ALGORITHMS TOUCH EVERYTHING: My wardrobe does not include a tinfoil hat, but I am very troubled at how much algorithms control what I see. Facebook, for example, doesn't distribute a Page's post to everyone who likes that page. It uses a number of factors to determine how many people get a post. What are those factors? Facebook isn't tellin'.
 
Google controls what you see in its search results, and in what order, with other algorithms. What factors are included? There's a lot of guessing, but nobody knows for sure but Google. 
 
I'm sure the rationale to keep these things secret is so that people can't game them. But there's SO much potential for abuse. Facebook has already admitted it manipulated user feeds for an experiment and research has suggested that social networks can fool people into thinking that things are common and popular when they're not.
 
Is it *that* big a step to imagine Facebook manipulating organic page reach in favor of a political candidate? Add to that the fact that Millenials use Facebook as a primary source of political news -  - and can you see how much power Facebook holds on a lot of fronts, using distribution algorithms that are not in the least bit transparent?
 
Q.  Do you remember a particular tip or strategy that you offered over the last few years that got lots of attention and comments? 
 
A. Two. 
 
In 2012 I created a Google Custom Search that searches official state, city, and county Web sites. Google had something similar but shut it down in 2011. You can read about it here.
 
In 2009 I wrote a quick article showing how to link to exact minutes and seconds in YouTube videos. That article remains ridiculously popular.  
 
Q. What do you think is a hot topic now in Web search that people are finding tricky and are struggling with and looking for advice?
 
A. The biggest one, the thing that drives me crazy, is that there isn't a good podcast search engine. Every time you turn around you'll read a story about how popular podcasting is and how it's starting a new renaissance in audio blah blah blah. But there's no good search engine for finding podcasts! iTunes podcast search has always been awful. Podcasts are hot, everybody's listening, and yet there's no decent search engine. Why? It is not "sexy" enough? The Yo app managed to raise $1.5 million in funding and nobody can be bothered to make a decent podcast search engine. Things baffle me sometimes. 
 
Q. What and who do you rely on as your own trusted sources for keeping up with trends and strategies in Web research and searching?
 
A. I use a combination of RSS feeds, social media monitoring (THANK YOU NUZZEL!) and Google Alerts to find the resources I cover. They're all important. And of course, ResearchBuzz readers send me tips when they find things, and I have friends on Facebook who tag me with resources. 
 
I gave up trying to keep up with EVERYTHING on the Internet around 1996. It wasn't possible then so it's absolutely not possible now. But I hope what I am finding, and I am linking to, and I am talking about, is helping somebody out there. That's what it's all about, isn't it? 

No doubt you are hardly shocked to be told that many online reviews you come across online are phony. And the issue is not a new one at all, as evidenced by the today's lawsuit filed by Amazon against over 1,000 users who received compensation for posting 5 star reviews for a variety of books and other products.


And despite the efforts of Amazon and others to prohibit and prevent the creation of fake reviews, it's unlikely--probably impossible in fact--to eliminate these entirely.


But everyone, and not just business researchers, has an interest in developing a level of information literacy to better understand how to spot reviews that are most likely to be authentic, and most likely to be a fake.  That's going to be the most sustainable and elegant solution to avoid getting fooled by fake reviewers.



Here are a variety of strategies you can take to help you better evaluate reviews from unknown persons on the Web:

• Note whether there are there lots of reviews for the
item or only a few. In general, the greater the number
of reviews, the less likely that a small number of phony
reviews can impact the aggregate score. That's why a site
like TripAdvisor, which receives and then averages many
thousands of reviews, is often so useful.

• Try to find out what interest—particularly financial—there
might be in creating fake reviews for that category and
adjust your skepticism meter accordingly. For example,
there is little financial interest in pumping up positive
ratings of, say, the tastiest recipes for homemade salsa,
but there certainly would be for competing hotel chains.

• If you are mostly interested in learning the pros and cons
of a reviewed item, consider ignoring the one- and five star
ratings, as these have a higher likelihood of being
phony, and read the threes and fours to get more nuanced
descriptions.

And when looking at any single review, ask yourself:

• Does the review site provide some kind of authority
badge, indicating that this review was written by a person
whose identity can be verified? If so, that’s a big plus. If
not, try to find out a bit about the person yourself, such
as where else this person has written reviews. Sometimes
you can click on a person’s name to find his or her other
reviews. Or, when it is worth your time and the reviewer
has an associated name or handle, you can search that
name online to try to find the other reviews yourself. If
you can discover more about a reviewer, try to address the
following questions:

• How long have they been writing reviews? Someone
who has been writing reviews for several years is
probably less likely to be a shill than a brand-new
reviewer.

•  What kinds of products do they review? The most
suspicious are people writing very positive (or negative)
reviews on a single product and doing so on multiple
review sites. One trustworthy type are reviewers who
focus on a specific niche (e.g., the person regularly
reviews all types of newly released cameras) and also
display a clear in-depth expertise on their subject and
share lots of nitty-gritty details. But someone who
reviews a wide range of products may be trustworthy
too, as it may indicate that they are not shilling for a
specific company. It’s a tough call!

•  Another potentially useful strategy is to seek out and
read other reviews by the same reviewer, in particular
to see if he has evaluated something you have
personally used and are familiar with. Does the analysis
confirm or is in synch with what you already know?

The above strategies--and others and useful analysis tools--are discussed in my upcoming book, Find it Fast: Extracting Expert Information from Social Networks, Big Data and More, to be published next month by Information Today Inc/CyberAge.

What are your strategies for evaluating reviews from unknown sources? Please share them here!

There's been a variety of articles and discussions recently about the expected eventual replacement of keyword searching as our means for finding answers and information, and the rise of the automated "Intelligent Personal Assistant" that can anticipate what we need and want and then even go ahead and complete the desired tasks we want to complete.

These assistants--which today are probably best represented by Apple's Siri,  Microsoft's Cortana and Google's Google Now--work by examining and crunching mountains of data about our own profile and data around us and engage in probabilistic analyses to make its predictions. That ingested data can include, for example, our past search history, our social media profile, our current location, our digital calendar, our past online preferences, and countless digital signals surrounding us embedded in objects.

There is even an excellent and compelling book on this development, written by Stefan Weitz, (Weitz had been  the director of Microsoft Search) titled Search: How the Data Explosion Makes Us Smarter. In the book, Weitz  goes into great detail on the promise of how what we mean by search will change based on this revolution in online searching. 

As part of a recent interview I conducted with Weitz in the September 2015 issue of The Information Advisor's Guide to Internet Research, we described the forecast capabilities of the personal intelligent assistant like this: 

So, say you are at a hotel and ready to go to a conference. Today you might go online and look up the location, then call for a cab, and make sure you’ve ordered breakfast via room service in enough time to get there for the first meeting. But in the new search environment, your digital assistant will have your favorite “on the road” room service breakfast ordered at the right time to your room, a cab called and ready, and a digital display of the address of the hotel for you to give the cab.

So for this week's blog I just want to pose a question: one can imagine a variety of consumer-oriented applications like the above for an intelligence personal assistant, but are there possibilities for business researchers? Could this be useful for finding information on companies, industries, new technologies, market research, patents/copyrights, trade, international statistical information and so on. If so, what would this look like?

I'd be grateful to hear from you on your speculations on what the potential is, if any, of the personal intelligent assistant for the business researcher.

Thoughts?

Readers--While up in lovely Prince Edward Island for about a month I am taking some summer time off from blogging...Check back here at the end of August!

A First Tuesday Profile

Welcome to the July 2015 “First Tuesday” feature—this month on a Wednesday due to the holidays!  On the first Tuesday of every month we feature a one-on-one discussion with a person behind one of the previous month’s Best of the Business Web e-letter selections whom we felt would be particularly interesting to talk to.

This month we have chosen Petr Knoth, the founder of CORE (COnnecting REpositories), a meta repository and search engine containing over 24 million journal articles and other academic and scholarly documents held in approximately 700 data repositories around the world, housed at the UK Open University.

We chose CORE as one of our June 2015 Best of the Business Web selections, describing it as "a real gold mine of a research site.”

The following is an edited summary of our Skype discussion with Petr.

Q. What is your background?

A. I am originally from the Czech Republic, with a background in machine learning and text mining and I’m currently a Senior Data Scientist at Mendeley,  and a research fellow at Open University.

When I joined Open University about 7 years ago, I was doing my PhD and I needed access to research papers but it was hard to get them as all the papers were behind a pay wall. But the open access movement was starting; though I also found that you could not access all these open access articles in one place. So I thought let’s see about how to get all this together.  So the idea of CORE was not a search engine first, just a way to connect these together.  

So I began developing the idea in November 2010 and our first funding was obtained in Feb 2011. I began developing CORE myself for the first few months, though now I have stepped back some from the day to day which is being handled by my colleague Lucas Anastasiou.   

Q. What were some of the barriers and challenges you faced in developing the project?

A. There were both technical and legal barriers. The biggest issue has been and continues to be that there is no single harmonized protocol that we can use to get the data from the institutional repositories and publisher system platform, so we need to create specific systems for each one.

Q. Do you need to set up formal agreements with each platform, or do you just send out a spider?

A. It all depends on each entity’s own terms and conditions. In some cases we need a formal agreement and we will comply. We are not pirates here. And we don’t actually send out a spider, we harvest content which is a different process, meaning we access the meta data from these repositories and from that meta data try to identify links to a full text PDF. 

Q. What would you say is unique about CORE vs. other open access repositories?

A. Well, one is our ability to provide fulltext when we can find it. Also, we are trying to help those who are mining text who need access to raw data to extract knowledge and then be able to do interesting things with that data. Previously, only publishers or databases with the data were the only ones who could exploit the data which I thought was unfair; so we serve the community so others can reuse the data set and do something cool with it.

We do also serve the general public when they are searching the Web for scholarly works, and we serve governmental, funding and other institutional bodies who need to see what is going on in scholarly research to help make decisions on where research is heading and which areas are good candidates for future funding.

Q. Is CORE particularly strong in covering certain disciplines or particular countries?

A. We are discipline agnostic, so I’d say the representation of works in various disciplines closely aligns to what’s out there and available in general. In terms of countries and languages, after English the next most represented language is Spanish. And as a UK aggregator, we do try to prioritize works from the UK. After that we focus on the US, and then the rest of the world.

Q. What is next for CORE?

A. We have a lot of projects in the works. One that we are putting a lot of attention into is to develop a cloud computing infrastructure that can be used by text miners so they don’t have to download the text or use API, but use the cloud to employ text mining algorithms to process the data. We want to encourage people to be able to get access to the data and do cool things with it.

_________________________________________________________________________________________________

Every first Tuesday of the month we conduct a one on one interview with someone behind one of the more interesting business research sites we highlighted from the month before. Due to the Fourth of July holidays, this month our First Tuesday will appear tomorrow, the 8th. You can look forward to an interview with Dr. Peter Knoth, the creator of the The COnnecting REpositories (CORE) project, hosted at The Knowledge Media Institute (KMi), a research centre at the Open University (OU) in London.

So as my favorite wizard said--come back tomorrow!

An article published in the New York Times last week, titled “What’s the Matter with Polling?” cited recent high profile election polls that turned out to be wrong, and identified a myriad of reasons why the polling business is in trouble. Although the article focused specifically on election polling, the fundamental problems the piece cited are applicable to virtually all types of surveys, including market research.

The difficulties are not actually all that new. It’s been at least 10, and probably more like 15 years where survey organizations had to deal with huge social and technological changes such as the growth of cell phones; declining response rates due to people’s busy lives and increased privacy concerns; non-scientific Web based surveys (you can read a spirited debate on the credibility of social surveys  in the July/August issue of  The Information Advisor's Guide to Internet Research, the journal that I co-edit), and the challenge of capturing valid data when measuring anything related to the fast moving digital world. But the article does speak to an acceleration of these trends.

But what worries me though is something else, something even bigger.

It seems that we are simply not trusting polls anymore. More people dismiss them out of hand as biased, meaningless, or gamed to achieve a pre-determined outcome. (You can get a sense of some of these attitudes by reading the comments section of the Times' article.)

Some of this cynicism is justified and healthy. But my concern is the knee jerk rejection of even well conducted surveys as if these were simply another “opinion” or just BS. This attitude can dovetail with the other distressing phenomenon whereby people dismiss science and data when the results do not agree with what they believe, expect, or want them to be.

I think this attitude may be particularly true for young people. The GenX and Millennial generations lived during a time when so many traditional institutions have shown themselves to be less trustworthy and thereby are now seen as less credible. Furthermore, younger people have grown up in a media environment when fact, opinion and increasingly, advertising, are mixed together and conflated.

But just as some reject evolution, human induced climate change, or the science showing no link between vaccinations and autism, and as others refuse to answer questions from the Bureau of the Census as a form of governmental ‘intrusion” of one’s privacy, the same cynical or uninformed attitudes appear to be seeping into a general rejection of all surveys and polls as well.

That is too bad of course, as well conducted surveys provide valuable snapshots on what we all or a designated sub-population believe, think, and act and can serve as one important input to guide decisions ranging from determining public policy to introducing a new product.

What to do?

I think it’s going to be up to the survey industry’s trade associations and specific polling and market research organizations to tackle these challenges head on. This means much more transparency and education in how surveys and polls are conducted and more focus on enhancing information literacy for the public. People need to understand matters like sample size, representativeness, and get a few basic primers on probability too—for example, helping the public understand what that "plus/minus confidence level" really means—and that forecasts do not mean that there is a 100% probability that the forecast event will occur.

From a business researcher’s—or any researcher’s perspective—this is one more meta problem going on with “what’s the matter” with polling.  For more information on how to best understand surveys and polls, I recommend the site: "Polling Fundamentals and Concepts: An overview for Journalists" created by the Journalists Resource site at Harvard University's Shorenstein Center on Media, Politics and Public Policy, and available here: