magpie.user.server.operations

Class SearchRunner

java.lang.Object

magpie.user.server.operations.SearchRunner

@Path(value="search")
public class SearchRunner
extends java.lang.Object

Perform searches that involve several different ML models.

Search is formatted as a JSON file with a format:

• datasetType: Type of dataset used to store initial entries
• entryGenerator: Definition of a BaseEntryGenerator
• steps: Steps in the analysis, which can be:
  • Model: Name of model to be run. Results are stored as a property
  • Filter: Name and options filter to be employed

  Each of these should be formatted as a JSON object, with keys type (string) and options (JSON object)

• entryRanker: Method used to rank entries
  • method: Method used to rank entries
  • number: Number of top entries to select
  • minimize: Boolean, whether to minimize or maximize the objective function
  • options: Array, options for the objective function

This generates an output JSON structure:

• chosenEntries: List of entries, as string
• data: Detailed data for each model run

Field Summary

Fields

Modifier and Type	Field and Description
protected java.util.Set<java.lang.String>	Models Set of models that have been run

Constructor Summary

Constructors

Constructor and Description
SearchRunner()

Method Summary

Modifier and Type	Method and Description
org.json.JSONObject	convertToJSON(MultiPropertyDataset data) Convert the dataset into a JSON object
MultiPropertyDataset	getEntries(org.json.JSONObject searchDefinition) Generate a list of entries
protected java.lang.String	performSearch(org.json.JSONObject searchDefinition) Function that prepares the actual search
void	runFilter(MultiPropertyDataset data, org.json.JSONObject filterOptions) Run a new model, add entries to dataset
MultiPropertyDataset	runModel(MultiPropertyDataset data, org.json.JSONObject options) Run a new model, add entries to dataset
void	runRanker(MultiPropertyDataset data, org.json.JSONObject rankerOptions) Run the entry ranker to get the top entries
java.lang.String	runSearch(java.lang.String searchForm)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

Models

protected java.util.Set<java.lang.String> Models

Set of models that have been run

Constructor Detail

SearchRunner

public SearchRunner()

Method Detail

runSearch

@POST
 @Produces(value="application/json")
public java.lang.String runSearch(@FormParam(value="search")
                                                                             java.lang.String searchForm)

performSearch

protected java.lang.String performSearch(org.json.JSONObject searchDefinition)

Function that prepares the actual search

Parameters:

searchDefinition - Search specification

Returns:

Result

convertToJSON

public org.json.JSONObject convertToJSON(MultiPropertyDataset data)

Convert the dataset into a JSON object

Parameters:

data - Dataset to be converted

Returns:

Output describing the selected entries

runRanker

public void runRanker(MultiPropertyDataset data,
                      org.json.JSONObject rankerOptions)

Run the entry ranker to get the top entries

Parameters:

data - [in/out] Dataset to be ranked, will contain only top entries after running

rankerOptions - Optiosn for the ranker

runFilter

public void runFilter(MultiPropertyDataset data,
                      org.json.JSONObject filterOptions)

Run a new model, add entries to dataset

Parameters:

data - Dataset of entries to run

filterOptions - Options for model

runModel

public MultiPropertyDataset runModel(MultiPropertyDataset data,
                                     org.json.JSONObject options)

Run a new model, add entries to dataset

Parameters:

data - Dataset of entries to run

options - Options for model. Should have 1 entry: model name

Returns:

Entries in a new dataset

getEntries

public MultiPropertyDataset getEntries(org.json.JSONObject searchDefinition)

Generate a list of entries

Parameters:

searchDefinition - Definition of search routine, input to the search routine

Returns:

Dataset containing entries