I met with John Lutz and Pat Dunae to discuss possibility of connecting GIS map of current legal
properties in Victoria city with the the historical datasets we have (tax rolls, business directories and censuses). If done, would allow users to

1) click on a location on a map and get the information from one or more of the datasets associated with that location for a specific year or for a range of years (as specified by user)

2) specify an attribute of properties (e.g. church/temple or saloon) or person (e.g. name or race) and highlight all the properties which match that attribute (e.g. are a church or are a saloon, or are owned by or inhabited by a person of specified name or race). This may be extended to use colour to highlight each instance of each value for an attribute (e.g. all churches purple, all industrial properties blue, all commercial properties green, all residential properties red etc.)

I'll look into the issues below and come up with an estimate of what type and amount of labour will be needed to create the back-end datasets, the front-end user-interface and the code to process queries and report results.

Issues:

There are problems with mapping the records in the GIS dataset to each of the old datasets:
Type 1: We don't yet know if the conventions used for naming legal description and street addresses in that dataset are consistent with those in each of the other datasets.
Type 2: change in legal and street descriptions
- due to subdivision of lots over time, many records in the GIS database map to a fraction of a lot in the old dataset
- due to amalgamation of lots over time, many records in the GIS database map to more than one lot in the old data set
- instances where two lots in the GIS correspond to three lots in the old data set or vice versa
- the GIS dataset is itself changing as legal properties are amalgamated and subdivided, we don't know what the implications are
- streets were renumbered in c1908, how to deal with that?

In addition, there are issues with the number and consistency of values used to describe attributes of properties and people. For example, the same person may be described as Chinese, Oriental, Mongloid or Asian and a given business may similarly be described in a range of ways. How do we normalize those kinds of things?

Finally, there may be instances where we make editorial decisions about attributes belonging to certain people or properties (e.g. street address not provided in data, but we can work out what it must have been, or attributes of children not provided but assumed to be inherited from parents). In those cases we may need to flag that the information has been deduced rather than obtained directly, and possibly a degree of confidence.

Background info:

Each record in the GIS database has fields for:
- unique Property ID (PID)
- legal description
- street address
- polygon co-ordinates (which are mappable to real GIS values).

Each record in the tax roll database includes fields for:
- surname of property owner
- legal description of property
- street address of property

Each record in the dominion census of 1891 included fields for:
- surname
- given name
- no street or legal address
Each record in the check census of 1891 includes fields for:
- surname
- given name
- street address
- # of white people, # of chinese or aboriginal people
Later dominion/federal censuses contain street address

Each record in the business directory includes fields for:
- surname of owner
- street address

The records in the GIS database map legal description, street address and polygon co-ordinates to each other. Each of the other datasets has a street address, possibly a legal description, and a surname of a person (owner, occupant, subject).