Wikidata:Usability and usefulness/Scenarios

Scenarios are typical situations and tasks we want to support. They may have (often substantial) non-general details, to make it clear what the situation and task can mean to a user. Otherwise, misunderstandings about "what we actually want" arise easily.

If you want to add a scenario, please keep in mind that it should be from the users perspective in an actual situation. Neither "As a user I want a function to generate a list of items" nor "A user who creates lists will like the autolistgenerator because it makes list generation easier" would be scenarios, since they lack the actual situation/context that makes clear why the user wants to generate such a list.

Onboarding and beginners

Add a new item from scratch

Story: As a (new) user I want to create a new item

Example users: All create new items, but this is most relevant for

• the active Wikipedia user,
• the occasional Wikidata user

Examples:

1. I looked at some items and found one on my hometown missing. Now I want to create an item on it.
2. My town has a new mayor; the mayor is not yet on Wikidata as a person (heard at Wikimania2017).
   1. First I find a city I know (e.g. New York), click at the mayor linked there (assuming it is a relatively often used, well modeled item). I keep this mayor-item open in a tab.
   2. I create a new item for my town’s mayor, trying to use the properties that the famous mayor has by looking at the previously opened item.
   3. I go to my town’s item and link the newly created mayor-item there.

Plausibility:

• We are sure it happens
• We don't have research on how/why/in which context it happens exactly. Some hints are given in the late 2017 editing study, though in the study, editing many items is far more frequent than just adding one.

Problem with current implementation:

• The needed "add" button is among the least visible elements on the page (see usability test)
• We don't provide any help on which properties and values should be there, basically the UI expects the user to somehow know how to model this.

What we already did for catering towards this scenario:

• Phabricator-Ticket: [Story] Useful empty state of item and property T76243

Supplement an existing item as a beginner

Story: As a (new) editor I saw that some information is missing and I want to add it.

Example users: Potentially all (?) but this is most relevant for

• the active Wikipedia user,
• the occasional Wikidata user

Examples:

1. A town is on Wikidata, but not in which municipality it is in. The user wants to add the municipality.

Problem with current implementation:

• It is hard to find out if the information is already there.
• People can't find "add statement" or actually add a value when they want to add a statement (Usability test 7/2017)
• It is hard to find out how the information should be modeled, particularly choosing the right properties and possible qualifiers. More experienced users may use a "famous" item of the same kind assuming it is modeled well or look at recommended properties on a WikiProject (e.g. this one for Paintings) At least one user from the late 2017 editing study described this method.

Advanced features

Add multiple, similar items manually

Story: As an editor, I have a large (10-500) catalog of similar data I want to put in to Wikidata.

Example users: Most relevant for

• the not-yet-defined GLAM user
• the occasional Wikidata user

Examples:

1. The user has information on about 20 paintings from a catalog of an exhibition. Most of the are painted in Oil (one is in watercolors), all have different titles (though 3 are "unnamed" no1, 2 and 3), all are from the same period.
2. The user has an overview of administration units of a country, e.g. who is head of that unit, how many people live there, which is the unit's capital etc.
3. The user wants to create a list of all main chapters of an organization on Wikipedia. He wants to add the 100+ chapters to Wikidata to be able to generate a list of them. The source data is the database of the organization.

Evidence: Three users described this need to us.

Plausibility: Estimated to be high: It is easy to imagine this, as well as similar tasks to be something our users need to do (view of the UX team)

Problem: Currently, the user needs to select each property and each value oneself, even though many of those are the same. This is demotivating. The user could use a tool but a) there might not be one and b) or hard to use.

What we already did: We tried to find an "excel to wikidata"-tool or a similar tool for doing mid-size edits/inputs to WD. We have a rough idea of the API etc. of Wikidata and some technical background, so we should be able to make sense of solutions even if slightly technical. However, after 20min search we did not find a good way forward. In conversations (at Wikimania 2017) I (Jan) got the impression that the knowledge about how to use the tools is passed on between people rather than learned on ones own.

Add multiple, similar items semi-automatically

Story: As an editor, I have a large (100-x) catalog of similar data I want to put in to Wikidata.

Example users: The active Wikidata editor

Examples:

1. A user wants to supplement the air date to all episodes of a TV show
   1. …looks as items that model "Air Date" for a TV show to replicate this structure for the newly created air date data they want to add
   2. …lists all shows of that series that lack the the property and value "Air Date" for air date (needs knowledge of a tool like Listeria).
   3. …imports this list in a spreadsheet
   4. …writes all the air dates from a web source data in the spreadsheet
   5. …uses string concatination and search/replace to create a quick statement-syntax text
   6. …uses the text to run quick statements

Evidence: High. Several users documented this in the very active editors user research 2017

Problem: A complex bricolage-y chain of tools is needed to do this efficiently

Adding a reference

Story: As an editor, I want to add a reference, so my data is not deleted and

Example users: Potentially all users, but most relevant for

• the not-yet-defined GLAM user (since references are important in their field)
• the Wikipedia user (since they voice the critique most often)

Examples:

1. A user sees a "imported from Wikipedia" reference on the value for "Population" on the item Berlin. The user knows that the city council publishes reports on this. They search the document on the city’s website, check the value and replace the old reference (Plausible, but not originating from research)
2. A user sees that a reference is missing on the speed of light. The user knows that the »Speed of light«-Article of Wikipedia has a section on measuring the speed. They go to the article and find "http://www.bipm.org/en/CGPM/db/17/1/". They add it as a reference for the speed of light being 299792458 m/s. (Plausible, but not originating from research)
3. A user wants to add female painters to Wikidata. The artist they add was born 1940 and died 2012. They add a reference to the death date from a major newspaper’s obituary , since the person could plausibly still alive (in contrast to a 16th century painter) and a wrongly asserted date of death can cause up trouble for community and the (alive!) person as well since the information can propagate to other wikis and personal assistance using Wikidata. Luckily the same article also gives the birth date, so it can serve as a reference there. (As observed in the late 2017 user research)

Evidence:

1. A frequent critique of Wikidata is the perceived lack of references
2. It turned out in our very active editors research that the value of references depends on the likelihood of the statement being controversial. Date of death for example is likely to be under much scrutiny, so references are added there. Something that might not get a reference is "The Simpsons is-an InstanceOf TV Series"

Plausibility:

• References have a high inherent value for the project

Problem:

• It is not clear how references should be added: As property/value pairs? As reference to an item representing the particular source?
  • In the very active editors research, URL:Link were most frequent as given reference, but this would need quantitative support.
• It is tedious to add references: To the data for Title, Author, … is annoying.
• There is a copy references gadget, but it is often broken (User Research)

What we already did:

• There is a citoid gadget, but it only covers some types of publications.

Features in relation to other projects

Creating a list on Wikipedia based on Wikidata data

Story: As a Wikipedia Editor, I want to…

Examples:

• The user wants to create a list of all main chapters of an organization on Wikipedia.

Example users: most relevant for

• Active Wikipedia editor

Evidence:

1. Described by one user at WikiCon

Plausibility:

• High

Problem:

• Currently there is Listeria

What we already did:

Linking an item to a database

Story: As …

Examples:

Example users: most relevant for

• the not-yet-defined GLAM user

Evidence:

Plausibility:

Problem:

What we already did: