Archive

Archive for February, 2014

Big Data For Statistical Agencies

February 13, 2014 Leave a comment

Our guest speaker,  Cavan Capps,  who is Big Data Lead services presented this  talk as part of the Program on Information Science Brown Bag Series.

Cavan Capps is the U.S. Census Bureau’s Lead on Big Data processing. In that role he is focusing on new Big Data sources for use in official statistics, best practice private sector processing techniques and software/hardware configurations that may be used to improve statistical processes and products. Previously, Mr. Capps initiated, designed and managed a multi-enterprise, fully distributed, statistical network called the DataWeb.

Capps provided the following summary of his talk.

Big Data provides both challenges and opportunities for the official statistical community.  The difficult issues of privacy, statistical reliability, and methodological transparency will need to be addressed in order to make full use of Big Data in the official statistical community.  Improvements in statistical coverage at small geographies, new statistical measures, more timely data at perhaps lower costs are the potential opportunities. This talk will provides an overview of some of the research being done by the Census Bureau as it explores the use of “Big Data” for statistical agency purposes.

And he has also described the U.S. census efforts to incorporate big data in in this article:  http://magazine.amstat.org/blog/2013/08/01/official-statistics/

What struck me most about Capps’ talk and the overall project is how many disciplines have to be mastered to find an optimal solution.

  • Deep social science knowledge (especially economics, sociology, psychology, political science) to design the right survey measures, and to come up with theoretically and substantively coherent alternative measures;
  • carefully designed machine learning algorithms are needed to extract actionable information from non-traditional data sources; 
  • advances in statistical methodology are needed to guide adaptive survey design; make reliable inferences over dynamic social networks; and to measure and correct for bias in measures generated from non-traditional data sources and non-probability samples;
  • large scale computing is needed to do this all in real time;
  • information privacy science is required to ensure the results that are released (at scale, and in real time) continue to maintain the public trust; And…
  • information science methodology is required to ensure the quality, versioning, authenticity, provenance and reliability that is expected of the US Census.

This is indeed a complex project. And also given the diversity of  areas implicated, a stimulating one — it has resonated with many different projects and conversations at MIT.

Categories: Uncategorized

When is “free to use” not free?

February 7, 2014 Leave a comment

When is “free to use” not free? … when it doesn’t establish  specific rights.

NISO offered a recent opportunity to comment on the draft recommendation on ‘Open Access Metadata and Indicators’. The following is my commentary on the draft recommendation. You may also be interested in reading the other commentaries on this draft.

Response to request for public comments on on ‘Open Access Metadata and Indicators’

Dr. Micah Altman
Director of Research; Head/Scientist, Program on Information Research — MIT Library, Massachusetts Institute of Technology
Non-Resident Senior Fellow, The Brookings Institution

Introduction

Thank you for the opportunity to respond to this report. Metadata and indicators for Open Access publications are an area in which standardization would benefit the scholarly community. As a practicing social scientist, a librarian, and as MIT’s representative to NISO, I have worked extensively with (and contributed to the development of) metadata schemas, open and closed licenses, and open access publications. My contribution is made with this perspective.

Comments

Scope

The scope of the work originally approved by NISO members was to develop metadata and visual indicators that would enable a user to determine whether a specific article is openly accessible, and what other rights are available. The current draft is more limited in in two ways: First, the draft does not address visual indicators. Second, the metadata field proposed in the draft signals only the license available without providing information on specific usage rights.

The first limitation is a pragmatic limitation of scope. Common, well-structure metadata is a pre-condition for systematic and reliable visual indicators. These indicators may be developed later by NISO or other communities.

The second limitation in scope is more fundamental and problematic. Metadata indicating licenses is less directly actionable.  A user can take direct actions (e.g., to read, mine, disseminate, or reuse content) based on knowledge of rights, but cannot take such actions knowing only the URI of the license — unless the user has independently determined what rights are associated with every license encountered. Moreover, different users may (correctly or incorrectly) interpret the same license as implicating different sets of rights. This creates both additional effort and risk for users, which greatly limits the potential value of the proposed practice.

The implicit justification for this limitation of scope is not clearly argued. However it seems to be based the claims, made on page 2, that “This is a contentious area where political views on modes of access lead to differing interpretations of what constitutes ‘open access’” and  “Considering the political, legal, and technical issues involved, the working group agreed that a simple approach for transmitting a minimal set of information would be preferred.”  [1]

This is a mistake, in my judgment, since contention (or simply uncertainty) over the definition of “open access” notwithstanding,  there is a well-established and well-defined set of core criteria that apply to open licenses; these include: the 10 criteria comprising the Open Source Initiatives Open Source Definition [2]; the 11 criteria comprising the Open Knowledge Foundation Open Definition (many of which are reused from the OSI criteria) [3]; and the four license properties defined by creative commons. [4] These criteria currently are readily applicable to dozens of existing independently-created open licenses [5], which have been applied to millions of works. Although these properties may not be comprehensive, nor is there a universal agreement over which of these properties constitute “open access”, any plausible definition of Open Access should include at least one of these properties. Thus a metadata schema that could be used to signal these properties is feasible, and could be reliably used to indicate useful rights. These right elements should be added to complement the proposed license reference tag, which could be used to indicate other rights not covered in the schema.

Design Issues

The proposed draft lists nine motivating use cases. The general selection of use cases appears appropriate. However, the definition of success for each use case is not clearly defined, making the claim that use cases are satisfied arguable. Moreover, the free_to_read element does adequately address the use cases to which it is a proposed solution. The free_to_read element is defined as meaning that “content can be read or viewed by any user without payment or authentication” (pg 4) the purpose is to  “provide a very simple indication of the status of the content without making statements about any additional re-use rights or restrictions.” No other formal definition of usage rights or conditions for this element is provided in the draft.

Under the stated definition, rights  other than reading could be curtailed in any variety of ways —  including for example a restriction on the right to review, criticize or comment upon the material.  Thus the rights implied by the free_to_read  element are less then the minimal criteria provided by any plausible open access license. Furthermore, this element is claimed to comprise part of the solution to compliance evaluation use cases (use cases 5.8 and 5.9 in the draft). It cannot support that purpose – compliance auditing relies upon well-defined criteria, and the free_to_read  definition is fatally ambiguous. The free_to_read element should be removed from the draft. It should be replaced by a metadata attribute indicating readability as defined by metadata indicating ‘access’ rights, as defined by the open criteria listed above. [6]

Technical Issues

  • Finally, there are a number of changes to the technical implementation proposed:
  • There should be a declared namespace for the proposed XML license elements, so they can be used in a structured way in XML documents without including them separately in multiple schema definitions
  •  Semantic markup (e.g. RDF) is required so that these elements may be used in non-XML metadata
  •  A schema should be supplied that formally and unambiguously defines: which elements and attributes are required, which are repeatable, what datatypes (e.g. date formats) are allowable, and any implicit default values
  • Make explicit that license_ref may  refer to waivers (e.g. CC0) as well as licenses.

NOTES

[1] The draft also raises the concern, on page 4, that “no legal team is going to agree to allow any specific use based on metadata unless they have agreed that the license allows.” The evidence for this assertion is unclear. However, even if true, it can easily be addressed by using the criteria above to design standard rights metadata profiles for each license to complement the metadata attributes associated with individual document. A legal team can then vet the profiles associated with a license, could certify a registry responsible for maintaining such profiles, or could agree to accept profiles available from the licenses authors.

[2] http://opensource.org/osd

[3] http://opendefinition.org/od/

[4] http://wiki.creativecommons.org/License_Properties

[5] http://en.wikipedia.org/wiki/Comparison_of_free_and_open-source_software_licenses

[6] Alternatively, if the definition of metadata elements is simply beyond the capacity of this project, free_to_use should simply be replaced by a license_ref instance that includes a URI to a well known license that  established free_to_read. The latter could even be designed for this purpose. This would at least remove the ambiguity of the free_to_read condition, and further simplify the schema.

Categories: Uncategorized