Points, Exact Location: Plotted Objects

Using grid coordinates or electronic “total station” enables recording the individual location of objects on the field. In that case, the location are represented as points in archeoViz (triplets of x, y and z coordinates).

Points, Vague Location: Spits, Buckets, Sieved Objects, and Recording Errors

However, it is also common that x, y, z, coordinates by object are not available, for different reasons due to:

• recording errors, loss of information, resulting in the need to replace one of several coordinates values by ranges (e.g., a X value is missing for an object and is replaced by the minimal and maximal X values of the square where this object was found);
• choice of method (e.g., excavation made and recorded using spits of arbitrary depth, sieved objects, etc.).

In all these cases, we have to deal with vague location, when objects cannot be located as points but are somewhere between ranges of coordinates. Vague location can concern one, two, or the three spatial dimensions (the x, y, z coordinates, respectively).

Note that this feature can also be used to deal with the imprecision of topographical instruments.

Lines

Lines are useful geometries for representing relationships. In archaeology, these can be refitting relationships between object fragments, orientation (fabric measurements, etc. Lines are generated from data loaded as refitting data, either from the “Data” tab, or with the refits.df parameter of the archeoViz() function.

Surfaces

Surfaces are useful geometries for representing ground levels, trenches, pits, etc. In archeoViz, this can be achieved by defining a subset of points summarising the desired surface, then displaying the convex hull of this subset.

Refittings

Refittings are usually recorded by archaeologists in two ways:

1. by sets of refitting objects: using a two columns table, where a row corresponds to an object. The first column stores the object’s unique id and the second column stores the id of the set of refitting objects this object belongs to.
2. by refitting relationships: using a two columns table, where a row corresponds to a relationship. The first column stores the first object’s unique id and the second column stores the second object’s unique id.

Although the second data structure is more accurate, the first is more commonly used.

archeoViz processes and represents the two data structures in two ways:

• sets of refitting objects must be described using a specific column in the objects.df table (e.g. object_refits) and are represented by the color of points in the plots (like any other variable);
• refitting relationships must be described using the refits.df table and are visualised as segments connecting the refitting objects in the plots.

Fabric Measurements

So far, archeoViz does not handle fabric measurements properly. However, the process used to represent refittings can also adapted and used to represent fabric measurements. This requires to distort the data structure as follows:

• assuming that fabric measurements were recorded with two measurements on each object (and not with dip and plunge measurements),
• different unique id values must be given to the two points, and
• the two measures are processed as if they were two refitting objects.

See an example of this method here.

Formatting Data

The tables must be CSV files with the first row containing the column labels. Contents in HTML are allowed. This makes it possible, in particular, to add links to external resources (e.g., to objects permanent identifier in other databases, or to concepts identifiers in standard ontologies / thesauri, etc.).

Formatting your data can be done:

• either using a spreadsheet editor on your machine to generate CSV files;
• or, for the objects table, using the SEAHORS application to load your data, define the variables (in the “Load data” tab), and export it to the archeoViz format (in the “Table” / “archeoViz exports” tab). It is also possible to directly send the data to an online archeoViz instance.

Objects Table

A row describes a single object with the following mandatory fields:

• id: alphanumerical value, unique identifier of the object
• xmin: numerical value, coordinate of the object on the X axis
• ymin: numerical value, coordinate of the object on the Y axis
• zmin: numerical value, coordinate of the object on the Z axis (positive depth value)
• layer: alphanumerical value, identifier of the object’s layer
• object_type: alphanumerical value, category of the object

In addition, optional fields are possible, including:

• square_x: alphanumerical value, identifier of the square on the X axis
• square_y: alphanumerical value, identifier of the square on the Y axis
• year : numerical value, year when the object was excavated
• xmax: numerical value, when the X location of the object is included in a range of X coordinates
• ymax: numerical value, when the Y location of the object is included in a range of Y coordinates
• zmax: numerical value, when the Z location of the object is included in a range of Z coordinates
• object_edit: unlimited number of additional variables describing the object (field names must start with object_ and have different suffixes)

The labels of the squares of the grid:

• are ordered alpha-numerically;
• are not displayed to avoid erroneous displays, if the number of labels does not correspond exactly to the total number of 100 (cm, m, km) squares that can be defined in the range of minimum and maximum coordinates contained in the xmin and ymin variables;
• can be completed with the add.x.square.labels and add.y.square.labels parameters of the archeoViz() function in order to add the missing labels (on the X and Y axes of the grid, respectively).

Refitting Table

A data table with two columns can be uploaded for refitting data (CSV format). Each row must contain the unique identifiers of two refitting objects (corresponding to the values of the id column in the objects table).

Timeline Table

A table (CSV format) can be uploaded about excavation history. Row gives the year when each grid square of the site was excavated or surveyed. This table must include the following variables:

• year: numerical value, the year of excavation
• square_x: alphanumerical value, identifier of the excavated square on the X axis
• square_y: alphanumerical value, identifier of the excavated square on the Y axis

Background Drawing

A background drawing can be displayed in the 3D and Map plots. This feature makes it possible, for example, to show a site map behind a point cloud. It requires a data table in which each row gives the X and Y coordinates of the points to be used to draw. Note that the lines will be drawn following the order of the points in the table. The coordinates system used must be the same that the one used for the objects. To draw several lines, an additional column (named “group”) is required, containing for each point the unique identifier of the line to which this point belongs to. The data set is loaded using the background.map parameter.

Units

By default, all distances in archeoViz are in centimeter. However, it can be changed by giving the unit parameter one of the following value: “cm”, “m”, “km”. This parameter determines the contents of the legend of the square size.

Through the Application Interface

The three types of tables can be loaded in the “Input data” tab. The CSV separator (one of: comma, semicolon, tabulation) and the character used for decimal points (period or comma).

Through URL parameters

The URL of an online instance of archeoViz can include the parameters:

• objects.df=
• refits.df=
• timeline.df=

whose values must be the URL of a CSV file observing the archeoViz format described above. For example: https://analytics.huma-num.fr/archeoviz/en/?objects.df=https://zenodo.org/record/8003880/files/bilzingsleben.csv

Generating Random Data

Using randomly generated data is made possible for demonstration purposes. To activate this feature, set the slider in “Input data” to a value higher than 0 (setting the value back to 0 deactivates the feature). An “objects” data set, a “refits” data set, and a “timeline” data set are generated, making it possible to test all the archeoViz functionalities.

Through the R function parameters

archeoViz’s launching function (archeoViz()) can be run without parameter

archeoViz()

or by using the objects.df, refits.df, or timeline.df parameters to input data.frames about the archaeological objects, refitting relationships between these objects, and the chronology of the excavation, respectively.

archeoViz(objects.df = NULL,  # data.frame with data about the objects
          refits.df = NULL,   # data.frame for refitting objects
          timeline.df = NULL) # optional data.frame for the excavation timeline

Rotating the points

You can change the orientation of the points in the plan. In the “Data” tab, select a value (in degrees) and click on the “Validate selection” button to confirm.

By Location Mode

If all the objects have exact location or vague location, then no option is proposed. However, if the dataset includes both exact and vague location, then it is possible to select only one or both of these options.

By Layer or Object Category

The data can be grouped in two ways: either based on their layer or on the selected “object_” variable. This option determines the colours of the points in the 3D and 2D plots and how to group points when computing convex hulls and 3D regression surfaces. Sub-sets can be defined by object categories, using the “variable” and “values” fields. Once one of the “object_type” (or other possible “object_” variables) is selected, its values appear below and can be selected using the tick boxes. The selection must be validated by clicking on the “Validate” button. This selection determines the data to be displayed in the plots and tables.

General features

The plots in the “3D plot”, “Map”, “Section X”, and “Section Y” tabs are generated using the plotly library. All the plots are dynamic and include a menu bar above the plot with several options (generating an image file, zooming, moving the view, etc). See details on the plotly website.

Clicking on a legend’s item modifies the display:

• a simple click on an item activates/deactivates its display;
• a double click on an item displays this item only (another double click cancels it).

This feature makes it possible to choose which layers are shown. In addition, the size of the points can be set and whether the refitting relationships must be represented or not.

Finally, clicking on a point displays information about that point in the table below the plot.

Visualising Spatial Uncertainty

In archeoViz, a distinction is made between exact location (given as x, y, z coordinates) and vague locations (given as ranges of coordinates). Both types of locations can be displayed. The uncertainty of vague locations can be visualised by representing objects not as points but as lines, planes, and volumes (if ranges of coordinates are given for one, two, or three spatial dimensions, respectively). Note that this feature is resource intensive and using it with too much data can significantly slow down the application.

Graphical outputs

Several graphical outputs can be generated in archeoViz.

• The plots in the “3D plot”, “Map”, “Section X”, and “Section Y” tabs can be exported:
  • in SVG format (by clicking on the “camera” icon in the menu bar above the plot),
  • as interactive plots in HTML format, by clicking on the “Export” button.
• The small map in the “Section X” and “Section Y” tabs can be exported in SVG by clicking on the “Download Map” link.
• The plan of the excavation chronology can be exported in SVG format by clicking on the “Download” button.

Regression surfaces

In the “3D plot” tab, clicking on “Compute surfaces” and “Validate” displays the regression surface associated with each layer (with at least 100 points). The surfaces are computed using the generalized additive model implemented in the mgcv package.

Convex hulls

In the “3D plot” tab, convex hulls can be displayed as follows:

1. tick the “Convex hulls” box,
2. select, from the menu that appears, the subsets of points for which convex hulls are to be calculated,
3. click on “Validate”.

Convex hulls associated with each subsets with at least 20 points are displayed. The convex hulls are computed using the cxhull package.

2D kernel density

In the “Map” tab, ticking the “Compute density” box and clicking on “Validate” generates a map with contour lines showing the points’ density. Density can be computed for all the points together or by layer (with at least 30 points). The 2D kernel density is computed with the kde2d function of the MASS package (through ggplot2).

Export from archeoViz

Data can be exported to other online applications from archeoViz “Statistics” tab. Some exports are possible only for specific types of data or if a minimum number of values is satisfied.

archeofrag is an R package and web application to assess and evaluate the distinctions betwen archaeological spatial units (e.g. layers) based on the analysis of refitting relationships between fragments of objects. The web version of the application includes methods to measure the cohesion and admixture of spatial units, and compare it to simulated data. If an instance of archeoViz is launched with refitting data, then this data can be analysed with archeofrag. See an example here.

The Seriograph is a web application (part of the SPARTAAS collection) to visualise changes in the quantitative distribution of artefacts types in ordered or unordered series of spatial units. Export to Seriograph is available from online archeoViz instance (only) for dataset with at least 2 different values for the layer variable and 2 different values for the selected variable (by default, object_type). See an example here.

Amado online is an on-line application for analysing contingency tables. Amado online allows you to manually reorder rows and columns, and perform automatic seriation and classification. Export to Amado online is available from online archeoViz instance (only) for dataset with at least 2 different values for the layer variable and 2 different values for the selected variable (by default, object_type). See an example here.

explor is an R Shiny / R package application for interactively exploring the results of multi-dimensional analyses. Export to explor is available from online archeoViz instance (only) for dataset with at least 2 different values for the layer variable and 2 different values for the selected variable (by default, object_type). The version of explor used from archeoViz is a fork of the original application, adapted to run correspondence analysis. See an example here.

shinyHeatmaply is an R Shiny / R package application to generate and interactively explore heatmaps. Multiple statistical distance and classification methods can be applied. Export to shinyHeatmaply is available from online archeoViz instance (only) for dataset with at least 2 different values for the layer variable and 2 different values for the selected variable (by default, object_type). The version of shinyHeatmaply used from archeoViz is a fork of the original application. See an example here.

Import to archeoViz

SEAHORS is a web application and R package to visualise the spatial distribution of archaeological remains. As mentioned above, SEAHORS can be used to import, reshape, and send a dataset to an online instance of the archeoViz application.

Square grid

archeoViz(square.size = 100, unit = "cm", rotation = 0, 
          grid.orientation = NULL, background.map = NULL,
          reverse.axis.values = NULL, reverse.square.names = NULL,
          add.x.square.labels = NULL, add.y.square.labels = NULL
          )

• square.size: numerical. Size (width and height) of the squares in the grid system. Default value is 100.
• unit: character. Unit for spatial distances. One of “cm”, “m”, “km”.
• rotation: integer. Value (degrees) for the in-plane rotation of the point cloud.
• grid.orientation: numerical. Orientation (degrees, positive or negative) of the grid (0 corresponds to a north orientation).
• background.map: data frame or matrix. Coordinates to draw background lines in 3D and Map plots.
• reverse.axis.values: character. Name of the axis or axes to be reversed (any combination of “x”, “y”, “z”).
• reverse.square.names: character. Name of the axis or axes for which to reverse the order of the square labels (any combination of “x”, “y”, “z”).
• add.x.square.labels: character. Additional square labels for the “x” axis.
• add.y.square.labels: character. Additional square labels for the “y” axis.

Parameter presetting

archeoViz(class.variable = NULL, class.values = NULL,
          default.group = "by.layer", location.mode = NULL,
          map.z.val = NULL, map.density = "no", map.refits = NULL,
          plot3d.hulls = NULL, plot3d.surfaces = NULL, plot3d.refits = NULL,
          sectionX.x.val = NULL, sectionX.y.val = NULL, sectionX.refits = NULL, 
          sectionY.x.val = NULL, sectionY.y.val = NULL, sectionY.refits = NULL,
          camera.center = NULL, camera.eye = NULL
          )

• class.variable: character. At the launch of the app, name of the variable to preselect.
• class.values: character vector. At the launch of the app, names of the values to preselect.
• default.group: character. At the launch of the app, preset of the variable used to group data (one of “by.layer” or “by.variable”).
• location.mode: character. At the launch of the app, preset of the location methods (any combination of “exact”, “fuzzy”, “show.uncertainty”).
• map.z.val: numerical. Minimal and maximal Z coordinates values to display in the map plot.
• map.density: character. At the launch of the app, whether to compute and show density contours in the map plot (one of “no”, “overall”, “by.variable”).
• map.refits: TRUE or FALSE. Whether to show refits in the map plot.
• plot3d.hulls: TRUE or FALSE. At the launch of the app, whether to compute and show convex hulls in the 3D plot.
• hulls.class.values: character. At the launch of the app, names of the points subsets for which to compute convex hulls.
• plot3d.surfaces: TRUE or FALSE. At the launch of the app, whether to compute and show regression in the 3D plot.
• plot3d.refits: TRUE or FALSE. At the launch of the app, whether to show refits on the 3D section plot.
• point.size: integer. At the launch of the app, size of the points in the plots.
• sectionX.x.val: numerical. At the launch of the app, minimal and maximal X coordinates values to display in the X section plot.
• sectionX.y.val: numerical. At the launch of the app, minimal and maximal Y coordinates values to display in the X section plot.
• sectionX.refits: TRUE or FALSE. At the launch of the app, whether to show refits in the X section plot.
• sectionY.x.val: numerical. At the launch of the app, minimal and maximal X coordinates values to display in the Y section plot.
• sectionY.y.val: numerical. At the launch of the app, minimal and maximal Y coordinates values to display in the Y section plot.
• sectionY.refits: TRUE or FALSE. At the launch of the app, whether to show refits in the Y section plot.
• camera.center: numerical. In 3D plot, coordinates of the point to which the camera looks at (default values: x=0, y=0, z=0).
• camera.eye: numerical. In 3D plot, coordinates of the camera’s position (default values: x=1.25, y=1.25, z=1.25).

Reactive plot display

archeoViz(run.plots = FALSE)

• run.plots: TRUE or FALSE. Whether to immediately compute and show plots (without requiring the user to click on the “Refresh” button).

Control Export Formats

• html.export : TRUE or FALSE. Whether or not to allow figures to be exported as interactive HTML widgets.
• table.export: TRUE or FALSE. Allow or disallow data transfer to third-party applications in the “Statistics” tab.

URL parameters

An instance of archeoViz deployed online on a server can be set with URL parameters. Supported parameters include:

• objects.df, refits.df, timeline.df
• title, home.text
• reverse.axis.values, reverse.square.names
• square.size
• add.x.square.labels, add.y.square.labels
• class.variable, class.values
• default.group
• location.mode
• map.density, map.refits
• plot3d.hulls, plot3d.surfaces, plot3d.refits
• sectionX.refits
• sectionY.refits
• run.plots

(The following parameters are not supported in the current version: map.z.val, sectionX.x.val, point.size, sectionX.y.val, sectionY.x.val, sectionY.y.val, lang, set.theme, camera.center, camera.eye, html.export, table.export.)

The parameters must be written using the URL syntax (?param1=value&param2=value2) and have the same type of values than when used in the R interface. For example, the following URL launches an archeoViz instance using the Bilzingsleben dataset:

https://analytics.huma-num.fr/archeoviz/en/?objects.df=https://zenodo.org/record/8003880/files/bilzingsleben.csv

This URL does the same, but also includes the refitting table (parameter &refits.df=) and set the activate the display of the refitting relationships in the 3D and map plots:

https://analytics.huma-num.fr/archeoviz/en/?map.refits=TRUE&plot3d.refits=TRUE&objects.df=https://zenodo.org/record/8003880/files/bilzingsleben.csv&refits.df=https://zenodo.org/record/8003880/files/bilzingsleben-antlers-refits.csv

The following URL launches the Bilzingsleben dataset, pre-setting the app to:

1. groups the points by variable (parameter default.group, with walue by.variable instead of by.layer)
2. selects only the “Antlers” (parameter class.values)
3. redefines the size of the square grid (parameter square.size, 500 instead of the 100 default value)
4. enable the immediate display of the plots (parameter run.plots)
5. modifies the title of the page (parameter title)
6. modifies the content of the home page with basic HTML contents (parameter home.txt)

https://analytics.huma-num.fr/archeoviz/en/?default.group=by.variable&class.values=Antler&square.size=500&run.plots=TRUE&title=Antlers%20at%20Bilzingsleben&home.text=Many%20antlers&objects.df=https://zenodo.org/record/8003880/files/bilzingsleben.csv

Note that the parameters add.x.square.labels, add.y.square.labels, location.mode, and class.values, which accept simple or multiple values in the R interface (e.g., c(“value1”, “value2”)) only accept one value when set as URL parameters (this is a restriction due to the URL syntax).