To Do

1. LATER Look up phpmyadmin for Joseph to use interacting with Database
2. DONE Organize the database
   • Worthy sent schema documents (or use UML-like syntax)
   • Build schema, then present to Worthy and Kathleen
   • Write script that copies data from the current organization (Jill’s DB) to our new schema
   • Update visualizations and api accordingly
3. DONE: Email Jill and ask what the problems were with the dates
   • Specifically note that not asking for work to be done, but asking for what happened so we can take care of it on our end.
4. DONE: Add duplicate field to person table (those that are duplicates will point to the original person)
5. Prepare abstract for University of Kansas Digital Humanities Forum: Deadline June 15
   • Provide a sentence on one or two directions the paper may go in along with the general information on the meeting
   • We have this marriage data, this humanities question, and we want to look at and analyze that data in cool ways (maybe include sankey or chord diagram, if possible). Kathleen would have to craft some of the humanities side of the abstract.

Interesting Names

• Mercy and Mary Fielding
• Robert Blashel Thompson (40865)
• Hyrum Smith (20563)
• Williard Richards
• Franklin D Richards (Son of Phineas Richards)
  • id: 468
  • Elizabeth Fate

Interesting Viz

• Brigham Young, Orson Spenser, and Albert Carrington, who’s children marry link
• Same as above, but including BY Jr link

Database Organization

Fields to consider about each person:

• Display Name
• Authoritative Name
• Alternate Name and Type w/ prefix and suffix
• Birth
• Birth Place
• Death
• Death Place
• Nonmember
• Baptism
• Baptism Place
• Baptism Officiator
• Baptism Proxy
• Male Priesthood Office
• Male Church Office
• Female Church Office
• RS date
• Washing and Anointing
• Endowment
• Marriage Time
• M Sealed Eternity
• M Sealed Time
• Officiator
• Proxy
• Child Natural
• Parent Natural Child
• Child Adopted - Multiple
• Second Anointing
• Divorce
• Cancellation
• Pioneer Company
• Pioneer Company Office
• Nauvoo Plat or Address
• Mission Date
• Mission Place
• Private Note Fields
• Private Note Fields
• Image
• Journal (T/F)

Also need to consider the following changes:

• Need alternate names
  • First, Middle, Last, Prefix, Suffix, Type
  • Type defines what kind of name it is:
    1. Maiden name
    2. Alternate Spelling
    3. Display Name
    4. Authoritative Name
  • Want a display name, authoritative name, and alternate names
• Separate events table that has significant life events
  • Could have just a baptism table
  • Data that would go inside this/these table(s)
    1. Baptisms: dates, officiator, participants, proxies, etc
    2. Marriages
    3. Sealings
• Always have note fields
  • Private Notes: note to self to add to the table
  • Public Notes: interesting notes that are available to the public
• Want to add a “Not a member of LDS” field
• Companies in crossing the plains
  • People had titles, this needs to be added to the entries
  • Who was in the office of the company? Who were the officers and what titles did they have?

Discussion

1. Sankey: Need to split people. For example, Mercy Fielding was married to Hyrum Smith and Robert Thompson during her lifetime.
2. Getting the Anointed Quorum together (Joseph is getting the list).

Publication Venues

University of Kansas Digital Humanities Forum: Deadline June 15

Meeting with Worthy, May 9, 2014

Questions/Discussion

1. Annual Progress Report: How much information to give? What should I discuss as topic?
2. Timing for summer, being out for next Thursday afternoon/Friday, if possible
3. Proposing: Possible over summer without taking credits, or need to wait until fall?
   • Committee: Glenn Wasson is still interested
4. Temporal Sets (Time-Dependent Sets)
   • New ideas for using timespans in the structure (insert for a given time span). Is the performance hit worth it?
   • See figure 1 example
5. Time-dependent Data Structures (generalized to temporal pointers with temporal aspects on the nodes)
   • Using partial computation
     • Degree (in and out) can be easily done by storing the degree at each node. Increment for next time is faster
     • Connectivity - can we denote connected components or the size thereof?
     • Farness (1/closeness) - can we store the sum of distance to all other nodes? How often need to recompute?
6. To Note: When visualizing social networks, knowing the future structure is very good (see BrInMa10.pdf, fig 2)
7. Some papers
   • See notes on website for 3 paper summaries

To Do

1. Look into Artificial Intelligence books. DONE.
   • Frame problem (when new knowledge comes in, change the known set of proofs/knowledge into another set)
   • Truth Maintenance Systems
2. Working sankey diagrams. DONE.
3. Finish the annual report. DONE.
   • Worthy asked Wes about the requirements in SIS
   • Goals: Propose early fall
   • Add research assistant with IATH and what I did here
4. Write algorithms and implement temporal sets
   • Marriages are perfect examples, see below
5. Build table structure for extended marriages (tribes?) so we don’t have to build them on the fly from the DB every time.
6. Add a sankey visualization with choices for church organization (currently Q12, 70) DONE.
7. Add time slider to sankey viz. THURS.
8. Directed graph viz with a person, the marriages they are linked to (adoption, birth, marriage) WED
9. Add levels of separation to sankey (0 is just the marriage requested, 1 would be their parents’ marriages and children’s marriages, etc) WED.
10. Fix out points so that if children get married, they go to the same dummy node at the end. DONE.

Discussion

1. Precompute structure
   • Build a table structure for the extended marriages (plural marriages) so that we don’t have to build every time directly from the database.
     • What is the order of marriages in these marriage units or tribes?
       • Is the order time dependent?
       • Are new root marriages formed when the original wife dies? divorces?
       • Do changes in root marriage change the order of the marriages?
       • Can a new wife be married in as root marriage, or will the next-in-line temporally take that position?
       • How do you get a new root marriage?

Meeting Notes and Questions, May 8, 2014

To Do

1. Figure out which day I can take off for the summer to do research. Thursday is best, or Thursday and Friday mornings before noon. DONE: Thursday.
2. Create a table view of women who are polyandrous (married multiple husbands). DONE.
3. Get real data into the Sankey diagram. DONE.
   • Limit to the Annointed Quorum (use the Q12 and Seventy until Kathleen gets back with the AQ list)
   • Have an option to ask for a specific person (such as Brigham Young) – later down the list, though
4. Reply to Jill re:Status of the date fix. Anything else I should do? DONE.
5. Look at wife death dates in the chord diagram over time. They seem to be sticking around after their deaths. DONE.
   • Ex: see Brigham Young and Miriam Works

Interesting Names

1. Susa Young (1856-1937)
   • Listed sometimes as the 13th apostle
   • Daughter of Brigham Young
2. _____ Diana Huntington
   • Married multiple times
3. Joseph Smith’s wives
   • Most polyandrous group
4. Willard Richards

Other Views

1. Choose a person and see the marriage units they are associated with. So, the biological parents of this person, but also any adoptions into other families. Show other relations between those marriage units as well, so noting if the person was adopted to a marriage unit because of their sister’s wedding into that marriage. See sketch on the notes from today.
2. Choose a church organization (AQ) and see all people in that group and their relations (how are they related by blood, office, sacrament, etc).
3. Quorum over time
   • Relationships between quorum and family
   • WE WANT TO SEE: “The frist time they take a second wife is X months after being annointed to the __ organization”
     • Is there a relationship between getting into the church hierarchy and starting to be polygamous, or vice versa?
   • Really cool idea for this graph is sketched on the second page of notes for today.

Organization of Relationships

This keeps changing, but we have

Church              Temple              Family
 - Authority         - Sacrimental       - Blood
 - Office

Also, a list of the relations:

• Sealing
• Blood
• Office (Priesthood) - Quorum (Annointed Quorum) - “Apostle”
• Sacrament (Temple)
• Ecclesiastical Position

Notes

• There were many women that had multiple husbands. It would be interesting to see how/if they were trading up to “better” men.

Sallaberry, et al 2012: Clustering, visualizing, and navigating for large dynmic graphs

They define a dynamic graph as an evolving graph where vertices and edges are added and removed over time. More formally, $G = (V,E)$, a dynamic graph, is an ordered swquence of subgraphs $S = \{ G_1 = (V_1, E_1), ..., (G_k = (V_k, E_k)\}$ where each $G_t$ is the subgraph of $G$ at time $t$. The edges and vertices of each subgraph are non-disjointed sets, with $V$ and $E$ being their unions. So, an edge can be in multiple subgraphs if it appears across multiple time points, and likewise for a vertex. What happens if a vertex changes value over time? Are they considered the same vertex? Can we still link the two vertices as the same?

Problems with previous approaches: stability of the layout. Previous work graphed each time step individually, then either displayed them as a movie or next to each other. These authors look at clusterings across time, and produce time-varying clusters that link clusters across time to better stabilize the cross-time display.

The authors define an ordering of clusters using a weighted quotient graph, where each node is a time-varying cluster of $G$ and edges exist between nodes if there are shared nodes in the clusters. Edge weights are the number of shared nodes. They define the Linear arrangement function $LA_\phi$ to find an ordering (see page 5).

A case study of autonomous systems on the internet among countries is used.

Without proving, they show their approach takes $\Theta(k*n)$ to draw a time-line ($k$ time steps, $n$ nodes), $\Theta(n)$ to layout a graph, and $\Theta(n + m)$ to render the graph ($m$ edges). However, there is “a long time” for preprocessing.

Brandes, et al 2011: Visualization Methods for Longitudinal Social Networks and Stochastic Actor-Oriented Modeling

Assumption: Longitudinal network data, a “time-ordered sequence of interrelated network observations that possibly differ in actor composition, structure, and attributes.” Their problem is defined as visualizing a given sequence of snapshots of an evolving social network, with intelligible connections between time slices (snapshots) so that the changes can be easily distinguished.

They consider only a node-link diagram, with actors as vertices and ties are edges. There are 4 objectives in layout (with reference, page 4):

1. Edges should be of more or less the same length
2. Vertices should be distributed well over the drawing area
3. The number of meaningless edge crossings should be kept small
4. Symmetries in graph structure should be visible in geometric symmetries

Specifically, the authors abandon the force-directed methods commonly used for the older stress minimization, or multidimensional scaling.

When drawing dynamic visualizations of social networks, knowledge of the future changes can aid in organizing the network layout. Can we store this information in our time-dependent data structures? Can we capture this change information and use it to display the future/past? The authors of this paper assume an offline model, so they know the future structure of the social network when trying to visualize the structure at any point in time.

Current proposals for addressing stability in dynamic network visualization are:

• Aggregation: all graphs in sequence are aggregated into one graph with one vertex for each actor (across time).
• Linking: all graphs in sequence are aggregated into one graph with one vertex for each actor at each time point, with all occurrences of each actor linked by edges.
• Anchoring: extra edges are used to anchor vertices to fixed positions (either a previous iteration of an actor or a reference point).

Case study: connections among 69 students at 15 intervals over a 1.5 year period.

Resources: Long list of social network visualization papers on page 2 (second paragraph).

Moody, et al 2005: Dynamic Network Visualization

The authors define two types of movies of social network visualization:

1. Flip books, where node positions remain constant, but edges change based on updated connections
2. Dynamic movies, where node positions move as relations change.

Addresses discrete versus continuous time. Discrete time consists of cross-sectional snapshots of the network, are cheaper and easier to implement. Continuous time consists of streaming relational events and interactions that have exact start/end points (ordered dyadic events). This is the kind of time we want to consider in both SNAC and the Mormon marriages. It’s possible to aggregate continuous time into discrete points, but not vice versa. The authors use a time window that spans a set of relational events into one aggregation of the network (note: this means that some lengths are incorrect–residues–since the continuous end-point may be at the beginning of the window, but the event has been considered to be occurred during the entire window). See figure 1, page 1213.

Social distance: the length of the shortest path in the network connecting two nodes.

Graph layout algorithms include

1. Force-directed or spring-embedder algorithms (common). Available in Pajek, NetMiner.
2. Multidimensional scaling (plot nodes using the dimensions that result from MDS techniques based on the geodesic distances or some measure of node similarities). Available in NetDraw, Krackplot, MultiNet. Reduce the $n-1$ dimensions present in a network to the 2 dimensions that capture most variance in the observed multidimensional distance.

Resources: Freeman 2000b, good review of history of network visualization.

Meeting Notes and Questions, May 1, 2014

Questions

1. What are the hierarchies in the church? The book pages are unclear as to how they should be grouped.
2. There are ambiguities in the database, is there enough information to sort out which are the people actually in those groups?
3. There are missing people, are they important enough and should we discuss this with Jill?
4. Marriage questions
   • What happens when a patriarch dies? Does the marriage disband or reorganize?
5. Updates on the Database
   • There are two organizations in the DB, Q12 and Seventy
   • Created table view of the organizations (see http://ford.cs.virginia.edu/nauvoo)

To Do

1. Get a list of people in the database that are in the AQ for Kathleen to vet. DONE.

Table alterations

Add the following to the Person table 1. Middle Name 2. Repeating Alternate Name field (AKA) 3. Repeating Title field 4. Preferred Name field

Quorum of the Annointed (1842 - Dec 1845)

• Includes Quorum of the 12 and others (First presidency, some of the Seventy)
• First endowment company (Collective exercies: go through the church/temple rites as companies)
• See emailed Document (Beki’s List for Access.pdf)
  • Group A came through in May 9th
  • Ignore Group C (for now)
• See emailed Document (AQ Working List.pdf)

Interesting Names

• Parley P. Pratt vs Parley Parker Pratt (id 434)
• Sidney Rigdon (id 42164)
• John Benett (sp) (id 17954 or 20690)
  • Defects in the summer chaos of 1842

Notes

See sketch of timeline on the notes for today.