Adds a new attribute for objects or links (depending on isObject)
that contains the IDs of the items in the corresponding table
Can be used to have qGraph conditions that limit the range of a vertex/edge
to a set of IDs in the database
Adds a new attribute for objects or links (depending on isObject)
that contains the IDs of the items in the corresponding table
Can be used to have qGraph conditions that limit the range of a vertex/edge
to a set of IDs in the database
Adds a random attribute between 0 and 1 to every object or link in the database --- serves the
same purpose as the above, but handles cases where ids are not sequential
Assigns a beta distributed value (weight) to each link
pdf: f(x) = k * x^(alpha-1) * (1-x)^(beta-1)
with k = g(alpha+beta)/(g(alpha)*g(beta)) and g(a) being the gamma function.
Assigns a uniformly distributed value (weight) to each link
pdf: f(x) = {0; for x < a, x>=b
{1/(b-a) for a<=xasteriskSource() -
Method in class kdl.prox.qgraph2.QGItem
To work around Sun bug #4654916 'JList and JTree should scroll automatically with first-letter navigation'
- http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4654916
From: http://groups.google.com/groups?hl=en&lr=&threadm=5p5s2m%2498b%40hershey.cc.gatech.edu&rnum=9&prev=/groups%3Fhl%3Den%26lr%3D%26q%3Djava%2Bdraw%2Barrow%26btnG%3DSearch
Computes an arrowhead polygon, located at the second point.
This class implements a Conditional Probability Distribution: a map between the different labels in the class,
and distributions of values for a given attribute.
Used by TFM 02 and TFM 12
A complicated routine that combines two consolitades vertices (represented
by the ObjTempSGINST tables)
Finds all links of type Y, As in vertex A, Bs in vertex B, and connects them,
Removes rows that do not match the Y annotation
and recomputes subg_id if there is an expansion because no annotation is present
Returns a table with the following format
a_(item_id subg_id name) o1_id o2_id link_id b_(item_id ...) new_subg_id
The resulting NST is materialized
A continuous distribution assumes that the values come from a continuous domain,
and uses a Guassian density function to estimate the probability of a given value
Copies the specified set of subgraph from copyFromCont into me
It recodes the original subg_ids so that they start from the current subg_id + 1
Copy involves two steps: copying objects/links and copying attributes.
corner -
Variable in class kdl.prox.gui2.rptviewer2.Scroll
CostComplexityPruning implements the pruning algorithm described in the monograph
Classification and Regression Trees by Breiman, Friedman, Olshen and Stone.
This is the constructor to use cross validation tree selection
Assumes the cross-validation folds are in respective train and test containers
nested within the main container e.g.,
mainContainer - train
- fold 0
- fold 1
...
A utility method that computes the feature vector from the aggr table with a threshold
(eg, count(gender='m') > 2), or avg(salary) > 10
Can be overloaded by aggregators that compute the feature vector in a different way (e.g., mode)
Creates a Monet ascii_io format file named fileName and containing the
column names and column types in colNamesAndTypes, separated by
COLUMN_DELIMITER.
Used by RPT.apply: creates tables for a list of already established FeatureSettings
NOTE: Remember to clear the nstCache before calling it --unless you intend to re-use old tables.
Recall behavior, from QGraphEditorJFrame:
o To create a subquery: Mouse-down on the background, drag to enclose the
desired vertices, and mouse-up to create the subquery around the surrounded
vertices.
Recall behavior, from QGraphEditorJFrame:
o To create a vertex: Click on the background (creates vertex in query) or
on a subquery (creates vertex in subquery).
The default statistic module allows uses to specify a burnIn time, to let the RDN move through
space before starting to record, and a skip time, to record every s iterations.
Creates a new atribute of type typeDef, and uses the dataNST passed in as
the storage of values (attrDataNST is not copied but used directly; it is made
persistent as well)
Delimits a string to match a particular dataTypeEnum:
'chr' -> delimit with single quotes
'date' -> delimit with date cast, with double quotes around value
'dbl' -> delimit with dbl cast
'oid' -> delimit with oid cast
'str' -> delimit with double quotes, and delmit embedded double quotes
with backslash ('\')
A null value is handled specially: It is cast to dataTypeEnum's type.
depth -
Variable in class kdl.prox.model2.rpt.RPTState
Title: Editor Query Iterator
Description: A query editor which completes subqueries before edges in order to properly
draw edges in yFiles which require all nodes to be present before any edges are
drawn.
This generation module looks at ALL the possible combinations of sources and aggregators by default, but
allows users to limit the list of aggregators via setValidAggregators.
Given an NST with items and an Attribute, it joins the NST with the tables for the
requested item and returns a new NST with columns for all the attrs.
Returns the name of bat where the data are stored
This might not be a real BAT, but instead a list of commands that
can be executed to get the values for the column
Returns the conditional distribution for a given source, once learned
Note that users are free to change this distribution --see, for example, LGM models
Copied from JEdit (http://cvs.sourceforge.net/viewcvs.py/jedit/jEdit/org/gjt/sp/jedit/MiscUtilities.java?rev=1.83&view=auto)
Returns the longest common prefix in the given set of strings.
Given a list of objects A, a list of objects B, and a set of links L
this method does two joins to compute
A X B, that is, an NST with the columns from A, L and B
and rows such that
A.item_id = L.o1_id && B.item_id = L.o2_id
(if isReverseDir, then A.item_id = L.o2_id && B.item_id = L.o1_id)
The join is done incrementally, with two calls to the NST join constructor
In the first pass, aObjs is joined with L (item_id = o1_id)
in the second pass, the result above is joined with B (o2_id = item_id)
The object lists may come as a consolidated vertex, with (item_id, subg_id, name)
or simply as a list of IDS (o_id).
Given a list of objects A, a list of objects B, and a set of links L
this method gets the list of connections, irrespective of the direction
of the links
It works by calling getObjectsConnectedViaDirectedLinks twice and then merging the results
The object lists may come as a consolidated vertex, with (item_id, subg_id, name)
or simply as a list of IDS (o_id).
Returns the value of the QUERY_ATTR_NAME attribute (used to store the query that generated the container)
Returns null if attribute doesn't exist, or isn't defined for this container
Given an NST with a set of observed values per subgraph, and a probability distribution which presumably saves
precomputed probabilities of those values, it returns a map probability> for each subgraph.
Return the name of the BATs that store the obj/link/attr
Returns a varialbe whose content is the name of that BAT (a str, not a bat)
The variable doesn't need to be release later.
The Graph interface is the lower-layer of the NSI package, encapsulating the access to the underlying database:
APP
------
SEARCH
------
NSI
------
GRAPH
All the Graph layer has to do is return a list of the neighbors of a particular node.
Adds a column at the end of the NST named newColName in which
different values correspond to different values in the specified columns
If the newColName column already exists, it deletes it.
G^2 statistic is defined as:
G^2 = 2 * \sum_{ijk} N_{ijk} ln [(N_{ijk} * N_{k})/(N_{ik} * N_{jk})]
where
N_{ijk} is the value stored at counts[i][j][k] and something like
N_{k} is equal to \sum_{ij} N_{ijk}.
Main application for the browser-style Proximity database browser user
interface (open databases, run scripts, browse attribute and collections,
etc.) Manages JFrames created by me so that we know when to exit.
Like the inMemory graph, but instead of setting the weight of the neighboring node to 1, it
actually uses the 'weight' link attribute to get the weight of the connection to the neighbors