October 28, 2010
ASIST conference 2010, Pittsburgh Day 3
FRBR Implementation and User Research
Studied mental models and used test group
Test group were asked to do a number of activities: card sorting, concept mapping, comparison of pairs
Concept mapping proved to be the most problematic- users had such drastically different ways at basic organization. Each one was different to line up the same set of concepts in a relational method. Some general tendencies, which did echo some of the FRBR structure
Kent State (w/ IMLS grant) studied practical applications. More project info here FRBR currently reflects the designers/researchers more so than the users
Used World Cat and the OCLC Fiction Finder for project (though noted that neither are strict uses of FRBR) Research included screen captures, eye tracking, think-aloud, survey interviews and focus groups. Found users did enjoyed the related items aspect of FRBR (likened it to Amazon)
WEM (Work-Expression-Manifestation) display
Still looking at the data analysis, so study is still on-going. Some confusion noted by users in the title field (Marc field 240), since this uniform title field can contain elements that users did not understand.
Imma Subirats from the Food and Agriculture Organization (FAO), United Nations
They looked into FRBR application for a number of reasons- the nature of their digital collection: 6 languages, content type (liked relational aspect for their research collection from conferences, etc). Fedora based. A reduced application of FRBR, in part due to legacy data and nature of their records (they do not use controlled names) Previous data also did not use LIS standards or guidelines (AGRO voc thesaurus
Merged records and converted into XML. Created relationships from the XML records. No live link- just about completed with project. Noted that Fedora made implementation rather easy.
Social Issues- Papers
Scholars who cite Twitter- Snowball sample, using 3 'core' academic researchers. Had to define the notion of citation for the project (direct link to peer-reviewed journal or a link to a blog that reviews journal) Looked at Qualitative and Quantitative data
Some scholars noted that they would follow certain colleagues/other professionals in the field on Twitter to act as a 'filter' to what they should be looking at (Alt-metrics)
Future research questions- What is the impact of these citations? Do all disciplines use this method? (Do some utilize social media better than others?)
Privacy Ramifications on Facebook
Katz v. U.S. case, 1967- An expectation of privacy
Is this expectation still reasonable in modern day?
We should have an awareness of privacy, and also what lines of privacy are drawn? Also questions 'digital memory'- Is your facebook profile ever really deleted?
Social Q+A- Impact of Text and Photo Cues
Tested people using dogster Looking at the difference in trust/reliability in the responses with text or text + photo. Ultimately, the photo did not add any reliability to this notion of truth, and rather the level of empathy in the text, and use of pronouns.
ASIST conference 2010, Pittsburgh Day 2
Role of Online Communities of Practice in Recent Responses to Disasters
How does technology enable the relief efforts after disaster? Research into twitter, wikis and blogs.
Dave Yates (University of Maryland, College Park) teaches, but is also an Air Force reserve. He was called on after the Haiti earthquake. Talked about the hierarchical response in the government, and how the State Department wiki proved useful in the mission (which has been non-classified since January). Maps could be updated and tagged with relevant info in a shared space easier than previous methods. A question was asked about the reliability of the information on this format, and Yates responded that a .gov email address was required for any changes to the wiki. So while the info on the wiki may not be 100% verified, it did prove to communicate important info between department more quickly. It also produced a rather lengthy wiki that is not the easiest to navigate. Yates pointed out that future efforts to coordinate info could be done in a better method (no one person was responsible for maintaining wiki content in this situation, for example)
Also mentioned individual and corporate efforts in social media after the Haiti disaster (text donations, Twitter, websites/blogs such as Le Projet Nouvelle Haiti which all asked for individual action)
In the Chinese earthquakes and floods in the past few years, the Chinese version of Twitter, Sina Weibo, has been studied recently by researchers looking at the types of tweets (Info-related, Opinion-related, Action-related or Emotion-related) and also polling the geography of the aftereffects of the earthquake. These tweets were tracked over time, and also compared to local and national news coverage trends, to see if there is any correlation between the two. Question from the audience to whether these tweets had a benefit to rescue efforts or other action; Some stories about family making connections through tweets; another of a civilian helping out a rescue effort by giving coordinates of a temporary landing spot for the relief helicopters.
Image Seeking, Retrieval and Use
A Comparison of a Conventional Taxonomy with a 3D Visualization for Use by Children
Moving 2D project into 3-D- Research project at McGill looking at children's search process within taxonomies. With input from children, the initial taxonomy was created, and later translated into a cone tree structure for 3-D browsing and searching. The terms were often 3-4 levels into the taxonomy, with varying levels of difficultly. A group of sixth graders were used to see how children would complete four searches. They compared these results with the 2D, and while the recall times were close, the kids enjoyed the interaction of the 3-D more.
ImageSieve: exploratory Search of Museum Archives with Named Entity-Based Faceted Browsing
Faceted browsing feature for image-based collections. Ran a study on people to see how they would use these different facets to assist in searching. The facets were divided through different tabs of the "who/what/where/when" categories. Also looked at the Flamenco project at UC Berkley. Flamenco (FLexible information Access using MEtadata in Novel COmbinations) uses keyword searches while also adhering to the structure of the collection structure. They could also tweak the system to reflect user searches (example of the proper citation of John F Kennedy, but also included JFK, Mr. President, Jack Kennedy)
Photo captions and descriptions were included in the search mechanism. They also used an element called "Shoebox" to let users click and drag certain objects over to build their own collection while they search and browse. Their collection for the project included 80,000 items, and also allowed users to annotate the images.
Investigating Variation in Querying Behavior for Image Search on the Web
How are users changing their query between a web-based search, and an image search? Main components of the project focused on task goals, working stage, search expertise and topic familiarity. The project rated the participant perception of the query results; in the satisfaction, confidence and usefulness of the result set. The length of the image searches were often much longer than that of text, though if an image was the focus of the search (vs. a web-page/text)
Linked data: Enabling Standards and other Approaches
Push to provide URIs/URLs/DOIs to every unique concept, digital object, etc to provide basis for linked data. Purpose to build relationships between all the various digital media with permanent, unique IDs. Extending work on OWL2 (Web Ontology Language): building URLs for people/place/things
Also refer to Networked planet on managing and publishing semantic data and Ontopia PSIs for unique subject IDs
Looking into issues of cross-referencing topics and drawing associations between concepts. Difficult with different terminologies. Ways to ID content include: ISBN, ISSN, ISTC (Text Code), ISRC (Recording Code), ISWC (Musical Work Code), ISAN (Audiovisual Number), ISN (Name Identifier), ISO 27729 (ISNI)
OCLC is acting as Assignment Agency for many of these conventions.
CISAC is acting as verification agency for music/arts)
SKOS RDF vocabs- a thesaurus used in a distributed, decentralized information system. Describing concepts and creating relationships between the concepts and terms (even the conceptual terms). Multi-lingual and extensible. "A single knowledge domain". A hierarchical and associative system.
RDFa (Resource Description Framework in Attributes) link to presentation slides
A way to assign rights and licensing info easily to digital material (Creative Commons example for embedded info) National Digital Newspaper project has RDFa embedded in every page to assure that rights info transfers with each page.
NISO I2 working group to assign institutional identifiers
-to establish provenance of digital information
-helping out the issue of digital authenticity and authoritative info
Still in final stages of development
Incubator group LLD (Library Linked Data) XG
Working to merge datasets (Linking Open Data) Problems include: licensing issues, mixed vocabularies, linking/interlinking material, quality of data; applying semantic wen technologies to Library Data (esp legacy data)
Working to ID collaboration tracks for future research; also how to build off existing initiatives; Need to find case studies
How to build a process from real world scenarios- how to build the tech pieces and requirements to fit a variety of data
LLD XG working with DCMI Architecture Forum Joint meeting info
October 25, 2010
ASIST conference 2010, Pittsburgh Day 1
Keynote Speaker: Lucy Suchman: Restoring Information's Body
I love when people from all types of study and specialties can come together to talk about shared dilemmas and issues & seem to actually be speaking the same language. Lucy Suchman is a professor of anthropology of science and technology at Lancaster University and also has twenty years experience as a researcher at Xerox's Research Center. While her talk did focus around her research in technology design and use, she spoke to larger issues of information access, use and structure. She brought up many underlying issues, How do we convey meaning in design?; How do we assign meaning and how do our respective roles/job titles/areas of expertise factor into our analysis of meaning? (She used an example of a photo- In one context, it was a woman holding a very particular group of flowers, while in another (w/out the caption or implied meaning, assumably), it was just a woman holding a bouquet of flowers. We provide the meaning, order and assign importance.
But what deems meaning of content? We can make independent inference of a thing on its own, but it is the associate, validated text that really gives meaning. Another independent application of another piece of knowledge, particularly in images. We provide order and meaning, but how do we create consistent dialogue? There seems to always be a level of human interaction and inference at any level (example of color matching of a soil sample in the sciences [Munsell color chart]- like relatively subjective, since color is relative to the beholder)
Technologies as "materialized figurations"
What is an object w/out meaning? (example of NRA slogan: guns don't kill people, people kill people) A gun in a drawer or a pocket is just a gun, but a gun in a hand implies intent. Suchman is very interested in the relationships of technology + human use.
Her more recent research is looking at both warfare and healthcare
Lucy's reading list:
How we became posthuman (N. Katherine Hayles)
Pandora's hope: essays on the reality of social science
Modest witness: Haraway
Acting in Anesthesia: Dawn Goodwin
Performance artist Stelarc- 2003 computer interactive likeness of artist- interacting through text between user and computer. Programmed to respond- Suchman asked computer about philosophy, life, poetry
Silenced work and invisible dialogues in knowledge representation: Susan Leigh Star
3:30 session- Knowledge Organization: Evaluating Foundation and Function in the Information Ecosystem
HIVE (Helping Interdisciplinary Vocabulary Engineering) s an IMLS funded project involving the Metadata Research Center and University of North Carolina at Chapel Hill. This is a way to integrate multiple controlled vocabularies. Since many disciplines require different ways of describing things (and will likely never merge into a singular "uber" list of terms, or at least anytime soon), there has to be a way to correlate terms and draw relations. Currently many of the widely known systems (LCSH, TGN, MeSH, etc) are in the project (10 total vocabs)
A concept browser has been experimented with to show related terms. Kea indexer being used in this project. Still needs research for front end usability and back-end performance issues.
Denise Bedford (World Bank/Kent State)- Has spent many years on the World Bank thesaurus- less on the mechanic end, but as a thesaurus creator. She sought out other organizations with similar projects to work collaboratively on these unique business geared models. (Noted that many of these projects at other organizations were eventually dropped due to the time and labor intensive work/lack of funds to sustain/database hard to manage after people retired, moved jobs, etc)
Talked about attempting to merge some of these efforts, but since each thesaurus is so specific to the organization, difficult to merge the efforts. She also noted that in her experience and research, that she feels that crosswalks rarely work, and are not worth the time. She has recently been looking at how to use these thesauri more effectively and how to increase the value of the work. She stressed the importance of the early meetings to negotiate meaning early on in the process, and not as the last step (lesson learned). She also stressed the importance of working at the entity level, and not the concept level, as tempting as it may be.
Bedfore is looking also at how these different systems/software can work together in a meaningful way. Also mentioned the Open Ontology Repository
Gail Hodge (Information International Associates, Inc. Trouble with terms with multiple meanings, esp. across disciplines (basic example of the word 'flow' in different sciences)
How does this compound with multiple languages?
Can we link the specific terms to the broader concepts successfully? (18 broad categories have already given some problems. How can we both offer the specificity for breadth while also catering to multiple disciplines in a single search mechanism or schema? (comment on whether we are looking at system-based problems or human based issues. Does a high level facet approach of an integrated modeling of database solve our issues? probably not- still boils down to the user and the purpose of search.
Should be looking to NISO, ISO and W3C for guidance