3. Data Abstraction

What?

Why do data semantics and types matter?

Data attributes define real-world meaning. (e.g. Basil, 7, S, Pear)

Type of the data: Structural interpretation

Data Level: Data Types

• item, attribute, link, position, grid

Dataset Level: Dataset Types

• How data types combined into larger structure
• table, tree, field or sampled value

Attribute Level: Attribute Types

• What kind of mathematical operations are meaningful?

→ Variable, dimension

---

Data Types

1. Attribute: Specific property that can be measured, observed, or logged. (variable or dimension)
2. Item: individual entity that is discrete
3. Link: relationship between items
4. Grid: specifies the strategy for sampling continuous data in terms of both geometric and topological relationships
5. Position: spatial data, providing a location

---

Dataset Types

Collection of info that is the target of analysis.

• Tables: Attributes+ Items
• Networks and Trees: Items + Links + Attributes
• Fields: Grids(positions) + Attributes
• Geometry: Items + Positions
• Clusters, sets, lists: Items

---

Tables

Items + Attributes (= field, variable, dimension)

Cell contains value

• quantitative

• nominal

• ordinal

• Multidimensional table: Indexed with multiple keys (cannot index with one key; SSN?)

Networks and Trees

Items + Attributes + Links

Well-suited when there is some relationship between items

• Node(item) → can have associated attributes
• Link : relation between items

Fields

Data created by sampling on grids

→ measurement of continuous domain

• Spatial Fields: cell structures based on sampling at spatial positions
  • non-spatial data: abstract data
  • Scivis: Spatial data is given with dataset (세포 위치가 정해져 있듯)
  • InfoVis: Designer가 space 사용에 대한 부분을 choose 한다.

(1) Spatial Fields: For SciVis. Spatial positions are given with the dataset.

(2) Grid Types: Grid needs (1) Geometry: location and (2) Topology: how each cell connected

• Uniform Grid
• Rectlinear grid (non-linear)
• Structured grid (curvlinear grid): geometry location should be specified
• Unstructured grid): topological and gemoetry should be informed

Geometry

• Specifies info about the shape of items, with explicit spatial positions (maps, geography)
• Often includes hierarchical structure (ex. 행정구역)
• GDP에 따른 나라 크기 distortion → 여전히 영국으로 인식되도록.

Combinations

set: unordered group of items

list: ordered items

cluster: group based on attribute similarity

path: ordered set + link

compound network: network with associated tree

---

Data Abstraction

• Domain specific to GENERIC
• translate domain-specific terms into words as generic as possible

Data Availability

• Static file: available at once
• Dynamic streams (ex. 실시간으로 변하는 Network)

---

Attribute Types

Categorical, ordered (ordinal, quantitative)

→ ordering direction (sequential, diverging, cyclic)

다른 책: Hierarchical attributes: within an attribute, or between multiple attributes (ex. ms, us, ns)

---

Levels of measurement

(1) Nominal: only named

• Name, categorical

(2) Ordinal: can be ordered

• Rank, Size

(3) Interval: distance is meaningful

• Temperature

(4) Ratio: Absolute zero

• Weight

---

Semantics (Key / Value)

for attributes

Key Attribute

index used to look up value attributes

• Implicit keys: keys are simply index in a row of the table
• Explicit keys: UUID, unique ordinal or categorical keys

In Fields… however

• Each cell represents continuous data distributed over domain.
• Defined as mapping from keys to values
• Multi variate structure depends on : # of value attribute
• Multi dimenstion structure depends on: # of key attribute

Scalar field, vector field, tensor field.

Semantic (Temporal)

• temporal attribute: related to time
• Complicated to handle
  • Multiscale: ns to decade
  • weeks don’t fit into months
• Time-varying semantics: time is key attribute.
  • not always spaced at uniform intervals! (응급실 등)
• Dynamic