Data Requests with an Agenda

This week I encountered an interesting example showing the problems inherent in sharing data with others who may have an agenda. Data is of no use to those who are not versed in the field they are examining, as we shall see.

In 2008 Richard E. Lenski (an evolutionary biologist) submitted a paper showing the result of a 20-year experiment that showed evolution in action: after more than 30,000 generations of 12 different populations of E. coli bacteria, one variant evolved the ability to use citrate (a component of the growth medium that can't usually be used as a carbon source by E. coli) as a carbon source. Andy Schafly (a Creationist) of the website Conservapedia sent a letter to Lenski asking for his data, quoting the submission guidelines for the Proceedings of the National Academy of Science that state that data must be made available to readers. To briefly summarize the dialog, Lenski's reply to Schafly's original request and followup made clear that Schafly had not read the article in-depth and did not have a background in evolutionary biology. Lenski pointed out the relevant methods and data were all shown in the paper, and offered to post data on three minor points which were left out of the paper (none of which concerned the existence of the citrate-using bacteria).

Lenski points out that Schafly seems to not be sure what he is asking for- his request for "data" is vague. One of Schafly's "acolytes" even states that Lenski should share the actual bacterial populations, since they form the basis of the data, as a way to "keep tax-payer-funded scientists honest". Lenski states that he will share the bacteria, but only with competent scientists, which Shafly clearly is not. A considerable amount of back and forth discussion has occured over this issue on blogs. (One notable post by a biophysics graduate student dissects Conservapedia's misinformed use of statistics.)

It is important to note that Lenski correctly identifies that Schafly is not asking for the data in good faith. The PNAS guidelines state that data sharing is for the purpose of "replicating and building upon work", while Schafly is only concerned with poking holes or disproving Lenski's results. Schafly takes a tactic often employed by those trying to promote the pseudoscience of "Intelligent Design" and attempts to poke holes anywhere he can in the paper while not being able to disprove the actual results. He even resorts to casting suspicion on the "astoundingly short 14-day peer review period", as if this were a valid argument.

We can see from this issue that making data freely available can have unexpected and troubling results. As has been mentioned in our readings, non-scientists without the skills to correctly interpret the data can draw incorrect conclusions, or may not even be able to recognize the data is staring them plain in the face within the paper. Schafly would probably not be satisfied even if the experiment was performed right in front of his face, even if he had 20 years to wait, but he seems to think that access to data means he should be allowed to see every piece of information generated in the experiment, down to lab notebooks that may only be comprehensible to Lenski and his assistants. Even when the raw data is presented in a table (as in the colony counts in Lenski's paper), those without scientific training may misuse the ideal of "data sharing" to attack scientists with baseless and misinformed accusations.

A new way to search?

Recently Nature.com implemented a new tool for searching the content of the Journal Nature called, appropriately, Nature.com Opensearch.
Over the course of the semester we have examined a many solutions and plans for storing data in repositories and in digital libraries, but the storage of information in these “data silos” is just one part of the data issue. Another piece of the puzzle is how data is found in a digital repository and this is the focus of natures new search strategy. What Nature.com has done is utilized the Opensearch specifications along with CQL to allow for the data in their collection to be searched freely by anyone who would like to build a widget to do so. Not a very revealing sentence I know, believe me I know, so let’s delve.
Opensearch is a set of specifications developed by Amazon (the geniuses who brought you the mechanical Turk) which are meant to allow information to be searched and used in ways that traditional search engines cannot.
Nature.com combined this set of specifications with the CQL, Contextual Query Language, which searches based on ‘semantic’ relationships rather than tradition syntax. This combined methodology offers several dimensions of flexibility to the system at Nature.com.
First the Open side of the application means that it can be embedded in a number of places (such as blogs etc) and results returned in any method that utilizes a standard XML format. The second and in my mind much more interesting aspect of this is that it allows a user to create specific search methodologies and returns. This page has several really nice examples.
Now what does this mean for us. Well in the end in this particular implementation is kind of limited in use. This is a pay Journal, so while anyone can see the articles that are returned not everyone will have access to the content of those article. Researchers with institutional access to the Journal will no doubt find this far more useful that the lay person. In the end what I am more interested in is the idea of turning the search function over to the general population. The institution in this case has set up a framework and turned access (read only, of course) over to anyone who might have a need for it. This way of searching where the data sets conform to standards and the applications which search them are fluid and changeable point to the promise that well curated digital collections can hold.

Microsoft's MyLifeBits: TOTAL DIGITAL CURATION

In the spirit of Microsoft trying to be absolutely everything to everyone, Jim Gemmell and Gordon Bell of the Microsoft Bay Area Research Center are in the process of developing an application that will archive literally every aspect of a person's life. The application is called "MyLifeBits." Well, it's not exactly an application yet, the project is still in the research phase. I heard about it while reading an article by Richard Cox of the School of Information Sciences at the University of Pittsburgh titled "Digital Curation and the Citizen Archivist." In the article, Cox discusses the role of the archivist in schooling the general public on strategies to preserve their digital artifacts, and he mentions the MyLifeBits project as a possible method to do just that.

The MyLifeBits project was inspired by Vannevar Bush's theoretical computer system - the Memex. In 1945, Bush developed a plan for a machine which would combine the functions of storage and electronic capture of images. This "Memex" (the word is a combination of "memory" and "index") would store a library of information on microfilm and all of the images would be connected by "associative links." The Memex has been credited as a pre-cursor to everything from the development of hypertext to the search engine, but in their paper "MyLifeBits: A Personal Database for Everything" Gemmell and Bell see themselves as fulfilling Bush's vision. They state that Bush "posited Memex as 'a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility. It is an enlarged intimate supplement to his memory."

Microsoft is taking the word "enlarged" seriously. Included in this application are the following tools (keep in mind that this is not an exhaustive list):

1. TV capture tool (to capture your TV watching)
   
2. SenseCam (to capture images IRL)
3. GPS import and Map display (to track your movements)
4. Radio capture
5. Telephone capture
6. IM capture
7. Browser tool (to track your web-browsing)
   
8. Legacy email client (you get the idea)
   
9. Voice annotation tool
10. Text annotation tool

Gemmell and Bell's argument for developing such an ambitious application with such wide-reaching curation tools is that since no one can predict when some item might be useful the "safest thing is to simply [sic] keep it all. Everything."

As I'm sure you can imagine, there are a whole host of issues in making such an enormous amount of information accessible, and so far it doesn't sound like Gemmell and Bell have developed a workable solution to these problems. They have been experimenting with hierarchical classification systems and have discovered (surprise!) that such systems quickly become unwieldy when dealing with large amounts of information. They haven't abandoned the idea of hierarchical folder systems, however, and are actually experimenting with "hierarchical classifications that will be developed by others to be downloaded by the user, and which contain extra information such as synonyms and descriptions to ease their use."

Classifications schemes are not the only thorny issue for the MyLifeBits team. Constructing useful metadata capture for video has also proved difficult. In developing a method to garner metadata for video they have suggested harvesting audio and converting it to searchable text so that, e.g., one could search for the name of a given person and jump to a segment of the video where the speaker mentions that person's name. Whether this would necessarily lead the searcher to a segment of the video with the person he/she is looking for is, well, who knows?

Clearly there are a number of issues that need to be ironed out before Microsoft can release this product, but Gemmell and Bell seem intent on doing so. Just this year, they released a book, Total Recall: How the E-Memory Revolution Will Change Everything, to proselytize their vision of total personal digital curation. While this project currently seems a little premature, it may be that in 30 years applications like MyLifeBits could be commonplace. If this is the direction that personal digital curation is going, what are some of the issues presented by such an all-encompassing view of "curation?" Certainly there are going to be copyright issues, classification issues, privacy issues (not to mention dystopian issues! Who wants their life to turn into some monitored, post-apocalyptic version of a Tom Cruise movie?)

Comparison of citation retrieval by search engines and the ISI Web of Knowledge

I continue to be curious about the practice of citation for promotion and tenure, and am always amazed when I haphazardly come across a print article that is so pertinent to our class discussion.

The September 2009 issue of College and Research Libraries (Vol. 70, No. 5, pp. 460-472) contains the article "A Citation Analysis of College & Research Libraries Comparing Yahoo, Google, Google Scholar, and ISI Web of Knowledge with Implications for Promotion and Tenure" by Charles Martell. (Web access is available to ALA members.)

1985 and 2005 studies determined that College & Research Libraries (C&RL) was ranked number one in "journal prestige in terms of value for tenure and promotion" by the Association of Research Libraries (ARL) Library Directors. Given this ranking, Martell determined the following: 1) frequency of citations of C&RL articles, 2) citation retrieval strengths of Yahoo, Google, Google Scholar, and ISI Web of Knowledge, 3) the relevance and applicability of these findings, and 4) classification and quantification of retrieved entries that did not qualify as citations. Every refereed C&RL article from 2000-2006 was used.

Advantages and drawbacks to each search method was discussed. Yahoo and Google are readily available and expansive, but laborious to filter. ISI WOK has a high level of quality and ease of use, but is fee-based and proprietary. Google Scholar has a broad coverage, but still has varying drawbacks: for example, for two of the top ten articles, over half the citations were from China. That said, Google Scholar returned more than twice as many citations as ISI WOK, and of course, the advantages and disadvantages should be weighed before selecting one over the other for citation counting.

It is valuable to note that this study was found to be particularly relevant to academic librarians in tenure-track positions. I found it curious that Martell discovered that deans of ALA-accredited education programs ranked C&RL 11th behind six information science journals and four library science journals. On second thought, perhaps not so curious given the trend of rebranding programs from "library science" to "information science" to incorporate the increasing importance of information technologies.

I think that this data would be very interesting to follow over the long term, in light of the growing interest in, or opposition to, open journals, and also our current discussion of broadening the definition of citations. It would be useful to note spikes/fluctuations in journal and/or citation usage and how it relates to the increasing effectiveness of search engines or initiatives such as the "Compact for Open Access Publishing Equity" (initiated by MIT, Cornell, Dartmouth, Harvard and UC Berkeley). If this data was made readily traceable, would it change the decisions and activities of the stakeholders (authors, promotion committees, publishers, etc.)?

What's the HapMap??

While clicking through the blogosphere earlier this week, I came across a project that seems to be an extension of the National Human Genome Research Institute. It's the International HapMap Project, with the intended focus to find "genes that affect health, disease, and individual responses to medications and environmental factors". The US, Canada, China, the UK, Nigeria, and Japan are collaborating on the research with public and private funds. Because it started in 2002, phase one of the project finished in 2005, with Phase II ending in 2007. I am interested in how accessible the data is and process of the project, which seems fairly documented.

The core of this project is to share information. The website says: "The International HapMap Project is not using the information in the HapMap to establish connections between particular genetic variants and diseases. Rather, the Project is designed to provide information that other researchers can use to link genetic variants to the risk for specific illnesses, which will lead to new methods of preventing, diagnosing, and treating disease."
Very commendable! It appears that this consortium is so expectant that their data will be used, that they offer tips on how to cite the project, especially how to refer to the sample populations (from Nigeria, Japan, Utah, etc). There is also a push to publish, as "researchers are encouraged to publish results based on combining HapMap data with data from other projects, particularly in efforts to find genes affecting a disease or a drug response. Researchers also are encouraged to use HapMap data to publish on the development of novel methods to analyze polymorphism, linkage disequilibrium, and association data."

Why all the good will? Is it because this is funded through the NIH? And must scholars who integrate the HapMap data with their own research make that data available? Interested in how much the data is being used, I looked at the site's publication page, but that hadn't been updated since 2007 when their last phase ended. So I went to GoogleScholar (gasp!) so I could just do a quick look up and found that the HapMap site has been cited 160 times. That could be good!

If I could even slightly understand anything about this project, I might be tempted to download the data (they provide tutorials!). Instead I just looked at the page from whence the files could come and was confused. There is a place to upload your own data, which I'm not sure I think is a particularly good idea, since their research seems quite controlled (270 participants only). I'm assuming the HapMap data and others' data is separate.

The International HapMap Project appears to be an example of a successful data-sharing hard science project, although its proximity in topic and funding agency to the Human Genome Project might also account for that. From an outsider's perspective, HapMap appears to have a socially conscious purpose (healing sick people, yay!) with an integrative approach to the data and publishing.

Challenges of Open-access Journals

An article titled "Ten Challenges for Open-access Journals" appeared in this month's issue of The Scholarly Publishing & Academic Resources Coalition (SPARC) Open Access Newsletter. In the article, Peter Suber explores what he sees as the ten greatest challenges facing open-acess journals.
Three of Suber's challenges deal with disparities between what is intended and what has been achieved, and the remaining seven have to do with doubts about open-access publications.
One of the disparities lies in the Impact Factors (IF) used to measure the impact of a particular journal. While IFs are intended to reveal the quality of the journal, a journal isn't even eligible for an IF until it has been around for 2 years. This is particularly hard on open-access journals because they are relatively new.
A second disparity is the failure of open-access journals to use a creative commons license to allow more than fair use of a journal. Many apen-access journals believe that by making their content free, they are truly open access. Suber argues that institutions are not free to exceed fair use. He suggests that until open-access journals explicitly allow for more uses (such as including the works in a database or archiving copies for preservation) they are not truly open-access.
The final disparity can be found in the gap between quality and prestige. It is impossible for a new journal to be prestigious from its start though it could be of a high quality. Subter argues that quality should matter more than prestige an until it does, open-access journals are at a disadvantage.
Subter moves on to discuss seven doubts that cause challenges for open-access publications. This include doubts about quality, preservation, honesty, publication fees, sustainability, redirection of funds, and even doubts about the strategies for addressing the challenges of open-access journals. These doubts are fairly self-explaintory and Subter proposes ways to deal with each of them. Subter suggests that an awareness and willingness to address these issues is absolutely necessary for the success of open-access journals.
We have previously discussed the need for changes in the publication process and even in the way an article's contribution is measured, but I still feel a bit doubtful that the change will happen anytime soon.

NSF's DataNet Partners so far: Data Conservancy and DataONE

Last spring, the Office of Cyberinfrastructure of the NSF announced its plan to support data preservation and access through the Sustainable Digital Data Preservation and Access Network Partners (DataNet) grant program. The program aims to address "one of the major challenges of this scientific generation: how to develop the new methods, management structures and technologies to manage the diversity, size, and complexity of current and future data sets and data streams." Through funds from the grant, "exemplar national and global data research infrastructure organizations," called DataNet Partners, will do just that. These organizations will combine library and archival sciences, cyberinfrastructure, computer and information sciences, and domain science expertise to provide preservation, access, integration and analysis capabilities over a decades-long timeline. As if that wasn't enough, these organizations should also be able to anticipate and adapt to changes in technologies and user expectations, stay cutting edge, and drive research and development. Oh, and one more thing: "potential applicants should note that this program is not intended to support narrowly-defined, discipline-specific repositories." The NSF is holding out for a hero!

Since the solicitation for proposals, two organizations have been awarded grants - both in August. (It seems that TeraGrid is categorized under the program but was awarded its $32 mil way back in '05.) $3.7 mil was given to Sayeed Choudhury of Johns Hopkins University in order to create the Data Conservancy (DC). Along with a long list of partners from universities and scientific organizations, Johns Hopkins will create "a new model in which libraries regard digital data as a special collection that must be maintained and served like their other collections." The Data Conservancy is concerned with the global carbon cycle and the carbon-climate-human system. The second award was given to William Michener of University of New Mexico and a ton of other partners including the CDL, Amazon, UIC, and Intel. That $12.3 million award is for the creation of DataONE (Observation Network for Earth). DataONE will address four challenges: preserving data, providing access to dispersed data, integrating data, and creating best practices for managing data. Its focus is huge: it will make "biological data available from the genome to the ecosystem; make environmental data available from atmospheric, ecological, hydrological, and oceanographic sources; provide secure and long-term preservation and access; and engage scientists, land-managers, policy makers, students, educators, and the public through logical access and intuitive visualizations."

It's interesting to me that both these DataNet Partners are doing pretty similar things, except the scope of DataONE seems to be much larger than that of the Data Conservancy. DataONE was presented at the 4th International Conference on Open Repositories at Georgia Tech in May. The presentation outlines some of the problems we've discussed with putting multi-disciplinary data in one place, but it purports that DataONE will "provide one-stop shopping for data" and will create "investigator toolkit" that will include nifty things like data visualization and Kepler workflow diagrams (remember those?) The presentation also gives an outline of how DataONE will be technologically implemented using the micro-services approach - or an "unbundled alternative to monolithic systems." Both Coudhury and Michener will speak at the next Educause conference in November, along with our pal Clifford Lynch, in a session called "Initiatives from the NSF's DataNet Program: DataONE and the Data Conservancy."

Sum of Humankind’s Knowledge Available Online

Great title for an article, huh? It appears that scientific research in Europe is becoming available to the public and other researchers through the DRIVER (Digital Repository Infrastructure Vision for European Research) Search Portal, which serves as a door to European Open Access research. The idea is to create a “library of libraries,” where individuals can readily search across repositories throughout Europe.

European researchers have created D-NET, software that links “information collected on diverse computer platforms, using legacy software which can still ‘talk’ or work with older systems in more than 25 European languages.” The software regularly harvests publications to provide access to information from numerous repositories. Visitors to the site are encouraged to use the software to set up other portals and are pointed toward examples of test cases where other groups are using it. The DRIVER portal site reports the following information about the collection:

• approximately 1,000,000 documents
• found in journal articles, dissertations, books, lectures, reports, etc.
• harvested regularly from more than 200 institutional or thematic repositories
• from 23 European countries
• in 25 languages.

The portal allows you to search broadly across repositories, but also limit your search by repository, language, document type, and date. At present the focus is on textual documents, but the idea is extend into other types of media. Current goals are to continue building the infrastructure and increasing the number of participating repositories. Another development of the initiative was the creation of Guidelines for Repository Managers, a document that outlines how to make content compatible with DRIVER. This type of document is an instructional tool to ease the process of increasing interoperability. The initiative is funded by the Research Infrastructure priority of the EU’s Sixth Framework Programme for research.

As a “library of libraries,” the DRIVER portal seems like a super repository that allows you to access a larger network of published materials through a single search, but I do not believe it goes beyond published works to data sets. However, the possibility for expansion into broader scholarly communication is there and the openness to using the software may allow for other uses within the scientific community.

Earth Simulator Center

Earth Simulator Center

In order to be able to visualize our readings for class this week, I selected one of the projects mentioned: the Japanese Earth Simulator Center and found it very interesting, innovative and a good example of using digital images to aid research.

The Earth Simulator Center (ESC) is part of the Yokohama Institute for Earth Sciences in Japan and is research center that appears to be a carbon copy of technologists and researchers working together create data. They research ocean temperatures, storm patterns, atmospheric conditions, and seem to track marine fluctuations. They take the data and construct different types of simulation diagrams. ESC states that their mission and statement for doing this is to "build a harmonius relationship between the Earth and human beings...through various areas of research and development" (Jamstec.g.jp).

The data gathered is used construct elaborate digital pictures of natural forces as they develop. There are pictures demonstrating geothermic changes with the presence of urban communities and without.

ESC also prides itself in working collaboratively, both on a domestic (in Japan) and on an international level with other research groups, mainly with marine scientists, and scientists that study clouds, weather, and other atmospheric forces, scientists that track changes on earth due to civilization or climate change. They have a feature on their website that lists all the projects that they're involved in with a link that either takes you another website or a webpage with their data on a pdf. However, most of the domestic project sites are written in Japanese and it is difficult for me ascertain what exactly the PDF reports could be.

I'm no scientist, but the pictures were easy for me to grasp the implications and meaning behind their constructions. They also exemplify a lot of qualities that are listed as progressive in building a science cyberinfrastructure community, such as collaboration, the cross disciplanary approach to research, and showing clearly how this technology can benefit scientific research.

SWORD2 and Infrastructure

This week I read JISC's SWORD2 Project Final Report [pdf] in conjunction with SWORD: Cutting Through the Red Tape to Populate Learning Materials Repositories, a February 2009 article introducing and explaining the protocol.

SWORD stands for Simple Web-service Offering Repository Deposit and is designed to "lower the barriers to deposit." It's on the default install of DSpace and is an optional interface for Fedora, EPrints, and IntraLibrary. Various repositories use the protocol and there are client-side applications that utilize it as well: Feedforward is a desktop application for personal information management, OfficeSWORD plugs into Microsoft Office so one could upload data to a repository directly from the program, and there's a widget and Facebook app for the protocol too.

The "Cutting Through the Red Tape" article briefly explains the underlying technology (it operates off of the Atom Publishing Protocol) and applications for use. It does this through four hypothetical use scenarios. The first details a university lecturer who can deposit notes, syllabi, transcripts and such simply by dragging them to an icon, and then a refinement he only needs to press a button in the application where he's creating the document (MS Office, say) and it's deposited. There's an optional form that could pop up for extra metadata he could attached (comments, etc.) but all the metadata the repository needs is already discernible (it's his work machine so they know who he is, time of deposit, format, size, title, and so on).

The second case examines two public sector educational resource providers (a social services one and a health one) that need to share data from their repositories. They're able to do this by using deposit services that use SWORD, so that every ingest with SWORD creates a set of metadata for discovery, search and display.

The third case elaborates upon direct deposit from the content creation tool, and the fourth demonstrates how one could receive various feeds from Feedforward, select a certain group, and deposit that into a repository.

The SWORD2 report essentially describes a successful interoperability update of the protocol and notes that advocacy efforts for SWORD should continue since it may be reaching "critical mass" and viability as a standard.

SWORD and the cases presented are interesting because I feel like it represents some of the infrastructure we've discussed. Particularly in the case OfficeSWORD, one could see how "invisible" it could become: just another button on the panel along with 'Print' and 'Save' that deposits the work for circulation in a repository. The cases where the user is able to deposit directly from the content creation tool are therefore really encouraging. It's obvious that a significant barrier to digital repositories is the click-count or other simple work that needs to be done for the deposit; if software or plugin developers only need to design a function for SWORD deposit, this could significantly change.

In terms of cyberinfrastructure, TCP/IP and HTTP are protocol and protocol sets that are so embedded in how "things just work" it's frequently difficult to imagine them actually being developed and adopted. No user ever has to deal with them. But it's clear we need similarly massively successful protocols fueling the cyberinfrastructure NSF calls for.

Science Big & Small

Salo, Dorothea. "Costs and Service Models for Data Curation." The Book of Trogool (Blog). Posted September 19, 2009. Available at http://scienceblogs.com/bookoftrogool/2009/09/cost_and_service_models_for_da.php (Accessed 10/06/2009).

This week I want to talk about a blog post (rather than an article) that caught my attention: Dorothea Salo's "Costs and Service Models for Data Curation." In her post, Salo discusses her concerns about the differences between how "Big Science" and "small science" will fare in matters of data curation and preservation. Perhaps because it is a blog post and not an article, Salo never quite defines "Big Science" and "small science," but from context and some reading around on the internet, it seems that "Big Science" generally refers to large multiple investigator projects, with big budgets, that tend to involve advanced (and expensive) technologies with large centralized labs (such as CERN), whereas "small science" involves a single investigator or a small team, usually working at a single lab on an independent research program whose funding is dependent on smaller limited-term grants. Salo sees the differences between big and small science as resulting in important disparities that affect data curation.

The problem as Salo, an academic librarian invested in e-research, sets it out is such: Big Science produces large, generally homogeneous data in huge quantities. However, because it's all part of one project, data standards evolve quickly and procedures can be institutionalized and set for the entire project. Thus once you get past the problem of storing huge amounts of data (and Salo is confident that they are coming up with solutions on this front), the human resources cost for curation is relatively small per terabyte of data. Salo characterizes small science, on the other hand, as tending to have ad hoc procedures without real data standards (since there are not enough people working on similar data who are willing to share that data and work together to generate standards). This is a problem because, without standards, each of small science's highly heterogeneous pieces of data will require "individual attention if it is to be adequately described and future-proofed." Accordingly, small science has the potential to have a frighteningly high human resources cost per terabyte.

This is significant because Salo intuits that small science, as a whole, has the potential to generate more research data than Big Science. It is also significant because, as she argues, all research data is potentially equally important since you can't know in advance where the groundbreaking, paradigm-altering insights will come from (especially since those breakthroughs may come from subsequent mash-ups of data). So, small science, which is least equipped to pay for long-term data curation, is likely to have the highest costs. This puts its "equally important" data at risk of getting lost in the data deluge (Salo explains in the subsequent comment posts that she has already seen instances of this type of loss). Salo is skeptical about the model of institutional cost-recovery cyberinfrastructure since she thinks Big Science will opt out (as she also explains in a subsequent comment) in favor of doing their own data curation with an embedded librarian whereas whatever money small science has to contribute may not be enough to adequately fund cost-recovery operations. The question that emerges for Salo is what type of business model will help us to address in a more equitable fashion the data curation disparities between Big and small science.

The comments that follow Salo's post also raise important questions. One of the respondents, Sayeed Choudhury, Director of digital curation at JHU (see Megan's email shout-out from earlier today), questions whether Big Science data really is as homogeneous as Salo presumes it to be. He also takes her to task in general (though, in the nicest way possible) for working from assumptions rather than concrete facts. He notes, for example that medical science is presumably "small science" but that the NIH has much more funding to distribute than the NSF. Choudhury suggests that libraries should become more involved in scientific data curation so that they can actually discover (rather than speculate about) what it entails in order to better build their infrastructures.

Choudhury's criticism that we need to actually compile and generate data on these issues is an important one. Salo's questions are provocative and, I think, important, which is why I've chosen to write about them here, but it is also unclear whether data curation in Big and small science really does work in the manner in which she supposes it does. Someone want to do a fact finding study?

In Atkins et al's NSF report "Revolutionizing Science and Engineering Through Cyberinfrastructure", the authors recognize that there is a "significant need [...] in many disciplines for long-term, distributed, and stable data and metadata repositories that institutionalize community data holdings" (42). In their imagining, these would provide tutorials on data formatting and quality control as well as offer tools for data preparation. This does seem like a potentially useful tactic for coping with data curation costs, but the question remains whether these tutorials and tools will really be equipped to cope with the heterogeneity of data. It also seems that these tools and standards would need to be engaged during the planning stages of researchers' projects if they wanted to avoid prohibitive human-resources costs further downstream. These are potentially important propositions, but we won't know how well such resources will serve small science until the tools are up and running and being widely used.

Cyberinfrastructurasaurus

This week I found an example of a scientist who is actively making use of cyberinfrastructure in his research. Allister Rees is a paleontologist/paleo-climatologist at the University of Arizona who coordinates the GEON PaleoIntegration Project (PIP). The PIP consists of a combination of five separate databases, available online, which can be applied individually or in any combination of the five at once. (The GEON PIP also features a free public portal, but you do have to request an account in order to have one authorized. What's really interesting is that the request form asks whether or not you want to be approved to post data to the databases, so there must be at least some data sharing going on. I did not request an account for fear of being an annoyance or time-waste, not being at all qualified to interpret paleontological data myself, but a preview tutorial is available here: PIP.) At any rate, the databases cover various eras of lithofacies (which I take to be referring to large patterns of sendimentary layers), the distribution of and data regarding certain types of climate sensitive rock and sediment from which the paleo-climate can be inferred, dinosaur distrabution, and plant fossil distrabution. The databases are then paired with paleographic maps and mapping tools.

In this brief article on the project, Rees asserts that his use of the PIP databases led him to conclude that dry, savanna-like climates tend to be the most likely to preserve dinosaur fossils, a conclusion which he claims Paleontologists had not yet realized since he was the first to directly look at paleo-vegetation diversity and the presence of dinosaur fossils in an overlay sort of function. Rees also feels like the PIP will be valuable in the study of mass extinctions and climate change.

I found the project interesting both because it illustrates the value for scientists in being able instantly to compare data from multiple specific fields courtesy of cyberinfrastructure and also because the presence of free, online databases does suggest that at least some scientists are willing to share their data.

NSF + COL = OOI (Ocean Observatories Initiative)

Once again the National Science Foundation is putting its hand into another effort to build scientific cyberinfrastructure, this time turning its attention to what lies under the sea. On Monday it was announced in a press release at the NSF website that the NSF and the Consortium for Ocean Leadership (COL) signed a cooperative agreement to build and operate the Ocean Observatories Initiative, which aims to “provide a network of undersea sensors” to observe the ocean and collect data like never before. From this article, it sounds like the NSF and the COL want to do for the seas something similar to what the National Virtual Observatory does for the sky.

This network of hundreds of sensors will be placed at several coastal, open-ocean, and seafloor locations, and will continuously collect data about “complex ocean processes” such as climate variability, ocean circulation, and ocean acidification. The Ocean Observatories Initiative looks to up the ante in data collection in terms of both rate and scale. And with this data, scientists studying the oceans hope to gain more of an understanding about how the oceans affect and interact with the earth and atmosphere.

This underwater initiative looks to take five or more years. The initial phase is the major construction phase, which includes “production engineering and prototyping of key coastal and open-ocean components (moorings, buoys, sensors), award of the primary seafloor cable contract, completion of a shore station for power and data, and software development for sensor interfaces to the network.” According to the article, the OOI should begin to see data flow by 2013, and the full system is projected to be up and running by 2015.

This is yet another effort that pours money into tools and technology to generate data, but the press release does not say much about what will be done with the data once it is collected. The article does say that the continuous data flow collected by the ocean sensors will be “integrated by a sophisticated computing network, and will be openly available to scientists, policy makers, students and the public.” Though the article fails to mention how exactly this will be made possible; I’m assuming that is because the project is still heavily in a development phase.

The OOI is a complex project, involving lots of cooperation from many different institutions, including Oregon State University, the Scripps Institution of Oceanography, the University of Washington, and University of California at San Diego (who is in charge of implementing the cyberinfrastructure component... though again the details are somewhat lacking).

So yet again the NSF is at the forefront of attempting to develop cyberinfrastructure, and once again a scientific field looks to be on the edge of a breakthrough when it comes to research and technology. It will be several years before we will start to really see the results of these efforts, and whether or not this project will be a success along the lines of the Virtual Observatory or another failed or incomplete attempt as we have encountered before in our readings. It sounds pretty exciting though, so I hope it works out.

Data-driven Scholarship in the Sciences and Humanities

In 2007, a piece entitled "The Virtual Observatory and the Roman de la Rose: Unexpected Relationships and the Collaborative Imperative" was posted on Academic Commons. It was authored by Sayeed Choudhury, the Director for Library Digital Programs and Hodson Director of the Digital Knowledge Center for the Sheridan Libraries at Johns Hopkins University, and Timothy Stinson, a post-doc fellow at Johns Hopkins with dual appointment in the Digital Research Curation Center and the Department of English. The article is concerned about the impact of cyberinfrastructure on the way that scholarship is conducted. The authors cite the Virtual Observatory and the Roman de la Rose Digital Library as two examples from different fields that exhibit the benefits afforded to scholarship thanks to cyberinfrastructure. The Virtual Observatory is a resource that brings together data from ground- and space-based telescopes, allowing scientists to find, retrieve, and analyze this astronomical data. The Roman de la Rose Digital Library, a collaborative effort between the Sheridan Libraries of Johns Hopkins University and the Bibliotheque national de France, is a digital library the goal of which is to create and house digital copies of all extant manuscripts of the Roman de la Rose.

Most would agree that the hard sciences are moving toward a data-driven model of scholarship, as evidenced by the Virtual Observatory project. However, the authors argue that the humanities, too, are beginning to be able to exploit (thanks to cyberinfrastructure) "data" for research in much that same way as the sciences. The parallel might at first seem not to fit, until one considers that data for the sciences and data for the humanities are two different things. For the sciences, data consist of sensor readings, scanner output, measurements of various kinds, and other things that are largely numerical in nature. However, humanities scholarship is based on a different, less straightforward type of data. As the authors of the Academic Commons article say, humanities materials can be considered data-rich in a variety of ways. For example, a single manuscript of a medieval text like the Roman de la Rose can contain many types of data in the form of illuminations, "artwork"/images, marginalia, annotations, as well as the semantic and linguistic data contained in the text itself. The creation of large digital libraries of this type is, for the humanities, analogous the creation of huge, complex datasets that make up digital curation projects like the Virtual Observatory, even though they seem comparatively data-poor.

The authors of this article further state that by exploiting resources such as the Roman de la Rose Digital Library, the humanities are poised to enter into the kind of collaborative scholarship that characterizes the hard sciences. "Digital tools," they assert, "are allowing us to capture, manipulate, and examine books and their data in ways that are revolutionizing the humanities." Digital versions of entire libraries can be created, and the contents of libraries that have been distributed over time can be virtually reassembled, creating new possibilities for scholarship. The authors recommend that humanities scholars work to define their own needs in the realm of cyberinfrastructure, and to create the specifications for and build the tools that they require for research.

The authors conclude by recognizing the imperative role that libraries and librarians will play in the future of these new data sets. While the role of the library in preserving the physical objects on which something like the Roman de la Rose Digital Library is based will remain vital, the library's role in preserving and making accessible the digital data that is created is one that is even more vital, all the more so because "the datasets are often not as highly regarded by libraries."

XML, JSON, and the importance of XSLT

This week Nature posted about their new search function meant for APIs. This new search interface will return results either using the tried and true (more later) XML or the newer and more streamlined JSON.

XML
I'm going to level with you guys. I secretly hate XML a little bit. When I heard about HTML5 and that XHTML was dead (sort of) I was psyched. Then I heard about JSON, and I had a bonafied I-told-you-so moment. (Quick summary so no panic ensues, HTML5 is combining many of the XHTML working groups deliverables into their own; you can still structure/nest tags neatly in HTML5, but it allows more flexibility; if there is one mistake on the page, the page won't fail.)

Then today I got the Flipsider email about the Federal Register in XML and I realized that XML isn't dead, even if aspects of it are going out of style. Not only are there many projects out there already using XML, but new ones will continue to pop up, no matter what w3c or JSON try.

In addition, XML is actually quite good at document mark-up (my fingers hurt just typing this.) I even like the strictness of syntax in XML (close inner tags then outer tags in the right order), but I think I (and, I venture to say, many information professionals) have confused XML as meant for something that it isn't meant for. XML is not a programming language. It is a document mark-up language that can be used to transport documents between programs. XML does not have a good way to deal with images or non-text data objects, although it claims to.

XML needs XSLT or other intermediary languages to be interpreted by browsers or other client-side applications.

JSON

JSON is a JavaScript data-interchange format. It is similar to XML in that it is meant as an envelope for items to be transported between programs, but it is simpler, and it was intended for use with programming languages. This makes it easier for use in browsers.

The best description of the difference between XML and JSON I found was at http://json.org/xml.html, "XML requires translating the structure of the data into a document structure. This mapping can be complicated. JSON structures are based on arrays and records. That is what data is made of. XML structures are based on elements (which can be nested), attributes (which cannot), raw content text, entities, DTDs, and other meta structures."

Both can be used, and XML can be transferred to JSON, the Nature approach is probably the best one.
Basically, I think I have to stop hating XML, but I can be grateful that there is something easier to program widgets with.

I apologize for the acronym party.

A Real Solution to High Textbook Prices

This week I read about a piece of legislation sure to be a hit with my fellow students: Dick Durbin (D-Ill.) has introduced the Open College Textbook Act, which would provide competitive grants to create open college textbooks. Although Durbin calls for the textbooks to be "freely" available, it seems like there may be a bit of confusion over whether these textbooks are open in the sense of open source software (free to modify and re-distribute) or whether they are truly free.

Durbin's plan avoids the pitfalls of Wikipedia and other similar open sites (which many college students use anyways) by providing for a review process of the material. In addition, Durbin is seeking to bypass the inflated prices we have seen recently by textbook publishers who bundle software and online portions to their textbooks. Personally, these addons often seemed like nothing more than an additional investment in time that students would have to take, instead of enrichment. Often, addons such as CD-ROMs are just ways for publishers to save money, by including PDF answer keys and similar content that could just as easily live online.

There are already free textbooks for basic subject such as physics and calculus online, freely accessible to students and the public alike. Since these subjects are well-established and not likely to change any time soon (you could just as easily teach yourself how to find derivatives with a textbook from the 1960s as from 2009) this is the right approach for these topics. Also, Durbin's approach follows other recent projects such as MIT's Open Courseware that make class materials available to anyone online. I can even think of one class here at the iSchool that has no textbook but instead pointers to free online tutorials that one can use to study HTML and PHP.

The question with these free content projects often falls down to "who's going to pay for it?" Since these are grant-funded textbooks, that problem has been solved, as long as the grants don't dry up. Universities seem to be of two minds on this topic: either they gladly make their course material available (as in the MIT case) or are worried about losing their intellectual property by putting the materials online. In addition, professors who have developed custom coursework often are in a bind because the university has rights to their own material, and they don't want to see it stolen and reused at other universities. Anyone whose professor has put passwords on downloadable course lecture notes will be familiar with this problem. This open solution could bypass the problem for professors: it's owned by no one (or everyone?) and is freely reusable.

The open access approach seems to be a nice step away from the consumer mentality that seems to plague current college students.Students "shop" for courses and professors and evaluate courses as if they are consumer products. See this article by Leah Schweitzer for a better analysis of this subject. In a way, education is a high-priced consumer good, but it is the work that students put in, in collaboration with professors, TAs and fellow students that can raise it above a passive experience. We want to believe that the halls of academia are more noble and lofty than the ITT Tech outlet in the local strip mall. Now, students with enough determination could in fact bypass the university and teach themselves with open course material.

I think this bill is a step in the right direction. Universities would have the chance to distinguish themselves in the level of personal service and creative projects they provide to students while still using the same materials as other schools. If academic papers and basic encyclopedias are available through database subscriptions at almost every college, there's no reason why basic textbooks shouldn't be available to everyone online. It would definitely make looking up the definition of Simple Harmonic Motion at 3AM easier right before that exam, without giving blind trust to Wikipedia.

PLoS and the new Metrics

This week I am looking at the Article Level Metrics recently implemented at PloS and what the might mean for the journal publishing world. One of the sublime pleasures of closely watching a growing field is how quickly the terrain can change.

In the tradition model of academic publishing the key metric to a “successful” article was the number of times it was cited in the literature. It should be noted here what a very poor metric this is to judge the influence of an article, the information that is contained in the publication could be well, suspect but if it provokes conversation or is used as a counter example then it will still be cited and still be seen as influential when counting citations. While perhaps there is no system that is perfect it is obvious that this traditional model could use a bit of up-dating.

The folks over at PloS have implemented a new set of article level metrics that go beyond the tradition of counting only citations and have begun counting other ways that articles are accessed and used.

The new metrics count the number of times the article is access, book marks, comments, ratings blog links as well as citation. The ability to collect all of this information is one of the exiting changes that online journals can offer. This collection of way that the journal is used can tell more about the quality of the work than the one dimensional view provided by examining citations alone.

Of course there are some know issues already when it comes to using these new metrics. For the most part PloS is up front about the issues inherent in these metrics and lists the steps that they are taking to address them. We should remember that each of these metrics has strengths and weakness but taken as a whole they reinforce each other and potentially have the power to change the way journals are used.

Now if we can just change the tenure paradigm in tandem.

Government Data

Government data is hugely useful in lots of applications. Three examples are social science research, transparency in government, and an individual's decision about where to live. It appears government data is making its way to the web. Data.gov hosts datasets from many different federal agencies, such as the Environmental Protection Agency, the Bureau of Transportation Statistics, the Centers for Disease Control and Prevention, and the Department of Education. You can get the raw data - XML, CSV or text - or you can use (more or less) people-friendly tools from their tool catalog. The tool catalog is a collection of widgets and extraction tools that are supposed to make the data on the site useful. For example, the MyEnvironment Widget allows a user to enter her zip code and get all kinds of related environmental data on a single page, complete with pie charts, maps, and graphs. I discovered a landfill in my neighborhood that I had no idea about. Crap. Related to data.gov is DataMasher, which we looked at in class earlier in the semester. It takes government data and allows registered users to create "mashups" of that data. For example, there are mashups of Hate Crimes vs Population, State Education Spending vs SAT scores, and Federal Spending per Congressional Representative. There's definitely a push out there to go beyond making the data available, but also making it useful to developers and end users.

And it's not just the federal government that is putting its data online. San Francisco recently launched datasf.org, which has datasets related to elections, environment, geography, health, housing, public safety, public works, and transportation. The site also features an App Showcase, which at this point mostly contains interactive maps and phone apps related to crime and the BART trains. I think it's a good idea to put the applications in one place; it makes it plainly obvious how important open data is. It occurs to me that if I were an app builder in Austin, I'd start pushing Austin to open up its data. Economic incentives might make the open data/transparent government pill a little easier to swallow.

The W3C, the definitive authority on all things web, published a working draft of Publishing Open Government Data in early September of 2009 that provides "straightforward steps to publishing government data." They recommend making the data available in well-known machine-readable formats such as XML, RDF, and CSV, while also making the data available in human-readable formats by putting it in HTML and using "permanent" and findable URIs. They provide guidance on documenting the data, preserving the data, not creating interfaces that prohibit access by other applications, and choosing what data to publish. The last point is an interesting one. How do you choose what data to publish? One of the people behind datasf.org, Jay Nath, writes about the 80-20 rule of government data. He discovered by analyzing downloads on the Washington DC Data Catalog that 80% of dataset downloads came from 20% of the datasets, mostly crime- and 311-related data. Nath and his group prioritized the data they made available on datasf.org based on these kinds of download statistics. One has to wonder, though, if the most unpopular datasets will ever make it online. If the only motivation for opening up government data is economic, then we might not ever see true transparency.

Cyberinfrastructure = Hardware + Software + Bandwidth + People

Cyberinfrastructure = Hardware + Software + Bandwidth + People

This week I was really not sure what I could write about, but then I came across Michael Roy's article in the Academic Commons that discussed the idea of cyberinfrastructure and building a computer research center at Northeastern University. I'm still working out my own ideas of what a cyberinfrastructure exactly is and this article featured a graspable array of issues and ideas that make up the necessity of establishing an infrastructure for research.

Roy begins his article by stating that computers were once thought to be for research, but then they became known to be teaching aids, but now the pendulum is swinging back to the idea that computers are used for research. He divides his article up into three parts that each address particular areas of establishing a IT research center: the practical issues, such as equipment, servers, staffing, the politics and the process and then the academic research process.

The first section elaborates on how Northeastern University decided to establish a Computer Research center. A committee was created and they interviewed the faculty to see what kind of research projects were currently being done and also asked about future projects to gain a picture of research needs. They decided at the end to build a cluster with features that focused on job management, size and interactvity. Then they had to set up a proposal to find the vendors that would be able to provide Northeastern with the technological equipment needed.

Needless to say, the IT Computer Research Center unleashed a series of critiques and skepticism amongst the faculty and within the academic world (mainly from the liberal arts world, predictably); however Leslie Hitch, Director of Academic Technology at Northeastern University, pointed out that Northeastern had a made an organizational goal to increase the quanity and quality of faculty research and the establishment of the Computing Research Center fell within the parameters of the university's mission. Then Roy began to discuss some of the challenges that they've had: staffing--how to staff the computing center, how much control to allow for, what defines productivity and so on.

Essentially, this article was useful for me to read because it provided some kind of framework of an individual institution's efforts to bank on digital research and clearly outlined some of the challenges we continue to have with defining the parameters of computing research and exactly what to do with it.

Twitter, Curation, and Horizontality

I'd like to start off by quoting The Boston Globe as reported by Tim Leberecht on The PopTech Blog: "Just as technology is giving us the ability to amplify every word we utter we have nothing really meaningful to say." While we could simply label this injunction as yet another irascible Bostonian diatribe, I think there is a good point here and it's not (of course) that no one is saying anything interesting anymore. The point is that people are saying SO MUCH that maybe we need some sort of filter or curator to let us know what is interesting and worthwhile and separate it from what is not. Does Twitter need a curator to make it meaningful?

We are already seeing some web-sites and applications which are doing exactly that. Breaking Tweets lists links to people are are tweeting about global "news worthy" events. Alltop lets you know about the celebrities and bloggers whom you "should" be following on Twitter. Applications like TweetDeck allow you to create your own filters so that you can organize your tweets and Facebook updates into categories like "news" and "friends" and "ridiculous" (there's a lot of that). From this, one can extrapolate ways in which other Twitter users could use websites to access tweets about various subjects: botanists could more easily access tweets about plant research, and journalists could access tweets about global events. Right now, the only way to filter tweets is by keyword searching for hash tags, but that can be a time-consuming and frustrating process.

While I can definitely appreciate the increased accessibility that such websites afford to Twitter users who have a specific interest, I also think there is something to be said for the horizontality of the current state of Twitter. All Twitter users have only 140 characters in which to express themselves, and no one is given more or less space based on their social status or another other factor. There's something egalitarian about the current state of Twitter and I think we're losing something important if we start imposing a hierarchical system of selection.

While I stated in last week's blog post that I like the idea of individual Twitter users curating web-content and posting it on Twitter, the idea of placing filters on Twitter itself seems to miss the point of Twitter altogether. In an interview for the Los Angeles Times, the co-creator of Twitter, Jack Dorsey, talks about the original concept of Twitter (this is a lengthy quote, but I feel like it's worth reproducing in its entirety): "The concept is so simple and so open-ended that people can make of it whatever they wish. They seek value and they add value. I've always said that Twitter is whatever you make of it. Because the first complaint we hear from everyone is: Why would I want to join this stupid useless thing and know what my brother's eating for lunch? But that really misses the point because Twitter is fundamentally recipient-controlled -- you choose to listen and you choose to leave. But you also choose what to put down and what to share. So if you decide to hook your plants up to Twitter and have it report when it needs to be watered, then that's a valid usage, or if you just decide to report what you're eating for lunch, that's a valid usage too." I think this gets at what I think is the uniqueness of Twitter - it's a giant behemoth of little snippets of information which the user must wade through to determine what he or she does or does not find interesting or useful. I don't know that we've really seen anything like this before, and I'm worried that if we start using websites that filter tweets for us, Twitter is going to become just another news website.

Twitter recently announced that it is, in fact, saving everyone's tweets. I like the idea of having one long, historical feed of 140 character blurbs with no form of organization apart from a date stamp. I think it will afford a unique, unfiltered way of looking at history.