Digital Odyssey 2013: Big Data, Small World Notes & Takeaways

Big Data

• 90% of the world’s data was created in the last 2 years
• can tell us much that other information cannot
• emphasize the need for analysis and interpretation
• your data is mined and used to make decisions for you, even more so in the future
• to prepare, know that big data will affect data management, discovery tools, new jobs, revised skills requirements, and revised infrastructures
• businesses will be made up of who has the most data and knows how to best use it

Data Visualization

Patrick Cain

• examples: homicide map, number of registered organ donours, bedbugs, drunk driving, crowd sourced neighbourhood maps
• providing information on what is otherwise a closed database
• problem is analyzing the data and drawing conclusions
• made people look at their community/neighbourhood in a different way
• more examples: includes information on free visualization tools
• a lot of public documents are behind a wall
• finding the balance between potential harm and making data open
• the data is imperfect, but help to tell a story
• have information on where the data comes from
• raw data gives you more power, can ask your own questions, but still have the difficulty of interpreting the data

The Changing Face of Toronto Public Library’s Data

Alan Harnum

• patrons do the usual: borrow, use ref, ask questions, attend programs, use computers, wireless, visit the website, etc.
• end up with a lot of transactional data, but it’s not all one big system, and not always available right away, in the form we want it, or at all
• conceptualization and modelling of data
• usage patterns are changing: digital borrow increasing, data needs are changing, policy needs are changing
• understanding data relationships: what influences what e.g. How many people use the library only for wireless?
• need for real-time data to improve service planning, responsiveness to inquiries, decision making
• policy implications
  • privacy – need to balance wanting to collect data with protecting customer privacy (datasets can be deanonymized)
  • service delivery – robust data, but without losing what patrons like
  • evaluation and measurement – how to use effectively while remembering that data is not the only decision making tool
• most important to remember: not everything that counts is counted, but not everything that is counted counts

Lunch Time

Lightning Talks

Big Data in Libraries

MJ Suhonos

• not actually as big in libraries: LoC: 1.9mill, Europeana: 20mill
• big data varies depending on the capacity
• think really complicated, but not actually that complex
• big data = cumbersome, out of our reach
• we don’t have to use the old tools, there are new ones
• have new opportunities
• cloud is not a magical bullet, just another tool – can do it in a more flexible way
• less about size and more about freedom and new opportunities because we didn’t have the tools in the past
• increasing the capacity around you
• can increase the discoverability of the long tail
• how to improve tools a little bit to solve problems we couldn’t before all over the place
• linked data is metadata infrastructure
• open data is policy infrastructure
• “The cloud is a lie”

Engagement and Impact of Twitter by Canadian Libraries

Angela Hamilton & Sarah Forbes

• found mostly analytic tools for marketers and for profit companies
• used tool to pull tweets: at 38k
• looking at the content, so no way to automate
• coding considerations: retweets, mentions, content type, tone, hashtags, links and media
• should have double coded to be more accurate
• should have gotten more help

On Dentographs

William Denton

• DDC in checkerboard to visualize depth and breath of the collection
• particularly good for collection comparisons
• LC in mountain version
• could do animation of how it changes year to year or day to day
• if use internal data, could do it based on circulation or holds
• by doing visualization, know what to do next time
• going from one medium to another, can extend
• Presentation Write up
• Code4Lib Journal Article

Open Data Policy in Canada

• Tracey Lauriault

examples: Open North, Hacking Health, Treaty Process, Residential School Map, municipal quality of life, AAAS remote sensing

where to get data

• Federal Data: opendata.gc.ca, research data Canada, data.gc.ca, Canadian International Development Agency
• Provinces: Ontario, Alberta, Quebec, BC, Saskatchewan
• Cities: 36 cities right now
• Community Data Portal

data advocacy

• Community Data Canada
• Canadian Council on Social Development
• Data Liberation Initiative

data policy

• not a lot of funding that requires data management
• Canadian Insititute of Health Research – encourage, but not policy
• Open Government Resolution by Office of the Information Commissioner of Canada – unfortunately don’t have a lot of power
• GeoConnections policy primers and guidelines
• Open Government Partnership – but only federal
• cippic – do everything in Canada
• should make public anything that is publicly funded

Privacy by Design: Big Privacy for Big Data

Michelle Chibba

• Ontario’s Information and Privacy Commissioner
• philosophy: consultation, cooperation, and collaboration
• confidentiality not the same as privacy
• privacy is all about the individual, and individual rights
• if unique, persistent, and linked to individual then it’s personally identifiable information (PII)
• people must be able to trust that organizations will manage their information properly
• forget the content, the metadata is what identifies the person
• most importantly, the information is persistent
• any digitization that can be intercepted and recreated into understandable information, then it is a record
• actually fairly easy to deanonymize information based on a few data points
• good data security is not the same thing as privacy
• most privacy breaches remain unknown
• need privacy by design
  • proactive
  • default
  • embedded
  • full functionality
  • security
  • viable and transparent
• can de-identify data using proper techniques based on the objectives/needs: Dispelling the myths surrounding de-identification
• data comanagement: accountability, minimization (collect little, use central registry), security, access
• UI design concepts tied to transparency and trust focused on context, awareness, discoverability, comprehension
• big data touching on privacy, example: connect un/structured data by casinos: card counter, relative of employee, or other relationships with the casino
• need to be able to use both unstructured and structured data to connect the dots
• features for next-generation sensemaking systems: full attribution, data tethering, analytics on anonymized data, tamper-resistent audit logs, false negative favouring methods, self-correction false positives, information transfer accounting
• your identity is your most valuable possession

Summary of Talk