org.terrier.querying
Class Manager

java.lang.Object
  org.terrier.querying.Manager

public class Manager
extends java.lang.Object

This class is responsible for handling/co-ordinating the main high-level operations of a query. These are:

Pre Processing (Term Pipeline, Control finding, term aggregration)

Matching

Post-processing @see org.terrier.querying.PostProcess

Post-filtering @see org.terrier.querying.PostFilter

</ul> Example usage:

 Manager m = new Manager(index);
 SearchRequest srq = m.newSearchRequest("Q1", "term1 title:term2");
 m.runPreProcessing(srq);
 m.runMatching(srq);
 m.runPostProcess(srq);
 m.runPostFilters(srq);
 

Properties

• querying.default.controls - sets the default controls for each query
• querying.allowed.controls - sets the controls which a users is allowed to set in a query
• querying.postprocesses.order - the order post processes should be run in
• querying.postprocesses.controls - mappings between controls and the post processes they should cause
• querying.preprocesses.order - the order pre processes should be run in
• querying.preprocesses.controls - mappings between controls and the pre processes they should cause
• querying.postfilters.order - the order post filters should be run in
• querying.postfilters.controls - mappings between controls and the post filters they should cause

Controls

• start : The result number to start at - defaults to 0 (1st result)
• end : the result number to end at - defaults to 0 (display all results)
• c : the c parameter for the DFR models, or more generally, the parameters for weighting model scheme

Field Summary

protected java.util.Set<java.lang.String>	Allowed_Controls This contains a list of controls that may be set in the querying API
protected java.util.Map<Index,java.util.Map<java.lang.String,Matching>>	Cache_Matching Cache loaded Matching models per Index in this map
protected java.util.Map<java.lang.String,PostFilter>	Cache_PostFilter Cache loaded PostFitler models in this map
protected java.util.Map<java.lang.String,PostProcess>	Cache_PostProcess Cache loaded PostProcess models in this map
protected java.util.Map<java.lang.String,Process>	Cache_PreProcess Cache loaded PostProcess models in this map
protected boolean	CACHING_FILTERS
protected java.util.Map<java.lang.String,java.lang.String>	Default_Controls This contains the mapping of controls and their values that should be set identially for each query created by this Manager
protected int	Defaults_Size How many default controls exist.
protected Index	index The index this querying comes from
protected static org.apache.log4j.Logger	logger
protected boolean	MATCH_EMPTY_QUERY
static java.lang.String	NAMESPACE_MATCHING The default namespace for Matching models to be loaded from
static java.lang.String	NAMESPACE_POSTFILTER The default namespace for PostFilters to be loaded from
static java.lang.String	NAMESPACE_POSTPROCESS The default namespace for PostProcesses to be loaded from
static java.lang.String	NAMESPACE_PREPROCESS The default namespace for PreProcesses to be loaded from
static java.lang.String	NAMESPACE_WEIGHTING The default namespace for Weighting models to be loaded from
protected java.lang.String	pipelineOutput This class is used as a TermPipelineAccessor, and this variable stores the result of the TermPipeline run for that term.
protected java.lang.String[][]	PostFilters_Controls A 2d array, contains (on 2nd level) the list of controls that turn on the PostFilters at the same 1st level place on PostFilters_Order
protected java.lang.String[]	PostFilters_Order An ordered list of post filters names.
protected java.lang.String[][]	PostProcesses_Controls A 2d array, contains (on 2nd level) the list of controls that turn on the PostProcesses at the same 1st level place on PostProcesses_Order
protected java.lang.String[]	PostProcesses_Order An ordered list of post process names.
protected java.lang.String[][]	PreProcesses_Controls A 2d array, contains (on 2nd level) the list of controls that turn on the PostProcesses at the same 1st level place on PostProcesses_Order
protected java.lang.String[]	PreProcesses_Order An ordered list of post process names.
protected static java.lang.String[][]	tinyDoubleStringArray
protected static java.lang.String[]	tinySingleStringArray
protected TermPipelineAccessor	tpa TermPipeline processing

Constructor Summary

Manager()
Default constructor.

Manager(Index _index)
Construct a Manager using the specified Index Throws IllegalArgumentException if the specified index is null

Method Summary

Index	getIndex() Returns the index used by the manager.
java.lang.String	getInfo(SearchRequest srq) Returns information about the weighting models and the post processors used for the given search request.
protected Matching	getMatchingModel(Request rq) Returns the matching model indicated to be used, based on the Index and the Matching name specified in the passed Request object.
protected PostFilter	getPostFilterModule(java.lang.String Name) Returns the post filter class named ModelName.
protected PostProcess	getPostProcessModule(java.lang.String Name) Returns the PostProcess named Name.
protected Process	getPreProcessModule(java.lang.String Name) Returns the PostProcess named Name.
protected Model	getWeightingModel(Request rq) Returns the weighting model requested by the Reqes from the WeightingModel factory.
protected void	load_controls_allowed() load in the controls that user is allowed to set
protected void	load_controls_default() load in the control defaults
protected void	load_pipeline() load in the term pipeline
protected void	load_postfilters_controls() load in the allowed post filter controls, and the order to run post processes in
protected void	load_postprocess_controls() load in the allowed postprocceses controls, and the order to run post processes in
protected void	load_preprocess_controls() load in the allowed postprocceses controls, and the order to run post processes in
SearchRequest	newSearchRequest() Ask for new SearchRequest object to be made.
SearchRequest	newSearchRequest(java.lang.String QueryID) Ask for new SearchRequest object to be made.
SearchRequest	newSearchRequest(java.lang.String QueryID, java.lang.String query) Ask for new SearchRequest object to be made, instantiated using the specified query id, and that the specified query should be parsed.
void	runMatching(SearchRequest srq) Runs the weighting and matching stage - this the main entry into the rest of the Terrier framework.
void	runPostFilters(SearchRequest srq) Runs the PostFilter modules in order added.
void	runPostProcessing(SearchRequest srq) Runs the PostProcessing modules in order added.
void	runPreProcessing(SearchRequest srq) runPreProcessing
protected void	setDefaults(Request srq) Set the default values for the controls of this new search request
void	setProperties(java.util.Properties p) Set all these properties.
void	setProperty(java.lang.String key, java.lang.String value) Provide a common interface for changing property values.
protected void	useThisIndex(Index i) use the index specified for the Manager

Methods inherited from class java.lang.Object

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

Field Detail

logger

protected static final org.apache.log4j.Logger logger

NAMESPACE_POSTPROCESS

public static final java.lang.String NAMESPACE_POSTPROCESS

The default namespace for PostProcesses to be loaded from

See Also:

Constant Field Values

NAMESPACE_PREPROCESS

public static final java.lang.String NAMESPACE_PREPROCESS

The default namespace for PreProcesses to be loaded from

See Also:

Constant Field Values

NAMESPACE_POSTFILTER

public static final java.lang.String NAMESPACE_POSTFILTER

The default namespace for PostFilters to be loaded from

See Also:

Constant Field Values

NAMESPACE_MATCHING

public static final java.lang.String NAMESPACE_MATCHING

The default namespace for Matching models to be loaded from

See Also:

Constant Field Values

NAMESPACE_WEIGHTING

public static final java.lang.String NAMESPACE_WEIGHTING

The default namespace for Weighting models to be loaded from

See Also:

Constant Field Values

Cache_Matching

protected java.util.Map<Index,java.util.Map<java.lang.String,Matching>> Cache_Matching

Cache loaded Matching models per Index in this map

Cache_PostProcess

protected java.util.Map<java.lang.String,PostProcess> Cache_PostProcess

Cache loaded PostProcess models in this map

Cache_PreProcess

protected java.util.Map<java.lang.String,Process> Cache_PreProcess

Cache loaded PostProcess models in this map

Cache_PostFilter

protected java.util.Map<java.lang.String,PostFilter> Cache_PostFilter

Cache loaded PostFitler models in this map

tpa

protected TermPipelineAccessor tpa

TermPipeline processing

index

protected Index index

The index this querying comes from

Allowed_Controls

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

This contains a list of controls that may be set in the querying API

Default_Controls

protected java.util.Map<java.lang.String,java.lang.String> Default_Controls

This contains the mapping of controls and their values that should be set identially for each query created by this Manager

Defaults_Size

protected int Defaults_Size

How many default controls exist. Directly corresponds to Default_Controls.size()

PostProcesses_Order

protected java.lang.String[] PostProcesses_Order

An ordered list of post process names. The controls at the same index in the PostProcesses_Controls list turn on the post process in this list.

PostProcesses_Controls

protected java.lang.String[][] PostProcesses_Controls

A 2d array, contains (on 2nd level) the list of controls that turn on the PostProcesses at the same 1st level place on PostProcesses_Order

PreProcesses_Order

protected java.lang.String[] PreProcesses_Order

An ordered list of post process names. The controls at the same index in the PostProcesses_Controls list turn on the post process in this list.

PreProcesses_Controls

protected java.lang.String[][] PreProcesses_Controls

A 2d array, contains (on 2nd level) the list of controls that turn on the PostProcesses at the same 1st level place on PostProcesses_Order

PostFilters_Order

protected java.lang.String[] PostFilters_Order

An ordered list of post filters names. The controls at the same index in the PostFilters_Controls list turn on the post process in this list.

PostFilters_Controls

protected java.lang.String[][] PostFilters_Controls

A 2d array, contains (on 2nd level) the list of controls that turn on the PostFilters at the same 1st level place on PostFilters_Order

pipelineOutput

protected java.lang.String pipelineOutput

This class is used as a TermPipelineAccessor, and this variable stores the result of the TermPipeline run for that term.

CACHING_FILTERS

protected boolean CACHING_FILTERS

MATCH_EMPTY_QUERY

protected final boolean MATCH_EMPTY_QUERY

tinySingleStringArray

protected static final java.lang.String[] tinySingleStringArray

tinyDoubleStringArray

protected static final java.lang.String[][] tinyDoubleStringArray

Constructor Detail

Manager

public Manager()

Default constructor. Use the default index

Since:

2.0

Manager

public Manager(Index _index)

Construct a Manager using the specified Index Throws IllegalArgumentException if the specified index is null

Parameters:

_index - The index to use in this manage

Method Detail

useThisIndex

protected void useThisIndex(Index i)

use the index specified for the Manager

load_controls_allowed

protected void load_controls_allowed()

load in the controls that user is allowed to set

load_controls_default

protected void load_controls_default()

load in the control defaults

load_postprocess_controls

protected void load_postprocess_controls()

load in the allowed postprocceses controls, and the order to run post processes in

load_preprocess_controls

protected void load_preprocess_controls()

load in the allowed postprocceses controls, and the order to run post processes in

load_postfilters_controls

protected void load_postfilters_controls()

load in the allowed post filter controls, and the order to run post processes in

load_pipeline

protected void load_pipeline()

load in the term pipeline

newSearchRequest

public SearchRequest newSearchRequest()

Ask for new SearchRequest object to be made. This is internally a Request object

newSearchRequest

public SearchRequest newSearchRequest(java.lang.String QueryID)

Ask for new SearchRequest object to be made. This is internally a Request object

Parameters:

QueryID - The request should be identified by QueryID

newSearchRequest

public SearchRequest newSearchRequest(java.lang.String QueryID,
                                      java.lang.String query)

Ask for new SearchRequest object to be made, instantiated using the specified query id, and that the specified query should be parsed.

Parameters:

QueryID - The request should be identified by QueryID

query - The actual user query

Returns:

The fully init'd search request for use in the manager

Since:

2.0

setDefaults

protected void setDefaults(Request srq)

Set the default values for the controls of this new search request

Parameters:

srq - The search request to have the default set to. This is done using the Default_Controls table, which is loaded by the load_controls_default method. The default are set in the properties file, by the querying.default.controls

getIndex

public Index getIndex()

Returns the index used by the manager. It is used for matching and possibly post-processing.

Returns:

Index the index used by the manager.

setProperty

public void setProperty(java.lang.String key,
                        java.lang.String value)

Provide a common interface for changing property values.

Parameters:

key - Key of property to set

value - Value of property to set

setProperties

public void setProperties(java.util.Properties p)

Set all these properties. Implemented using setProperty(String,String).

Parameters:

p - All properties to set

runPreProcessing

public void runPreProcessing(SearchRequest srq)

runPreProcessing

runMatching

public void runMatching(SearchRequest srq)

Runs the weighting and matching stage - this the main entry into the rest of the Terrier framework.

Parameters:

srq - the current SearchRequest object.

runPostProcessing

public void runPostProcessing(SearchRequest srq)

Runs the PostProcessing modules in order added. PostProcess modules alter the resultset. Examples might be query expansions which completely replaces the resultset.

Parameters:

srq - the current SearchRequest object.

runPostFilters

public void runPostFilters(SearchRequest srq)

Runs the PostFilter modules in order added. PostFilter modules filter the resultset. Examples might be removing results that don't have a hostname ending in the required postfix (site), or document ids that don't match a pattern (scope).

Parameters:

srq - the current SearchRequest object.

getMatchingModel

protected Matching getMatchingModel(Request rq)

Returns the matching model indicated to be used, based on the Index and the Matching name specified in the passed Request object. Caches already instantiaed matching models in Map Cache_Matching. If the matching model name doesn't contain '.', then NAMESPACE_MATCHING is prefixed to the name.

Parameters:

rq - The request indicating the Matching class, and the corresponding instance to use

Returns:

null If an error occurred obtaining the matching class

getWeightingModel

protected Model getWeightingModel(Request rq)

Returns the weighting model requested by the Reqes from the WeightingModel factory.

Parameters:

rq - The name of the weighting model to instantiate

getPostProcessModule

protected PostProcess getPostProcessModule(java.lang.String Name)

Returns the PostProcess named Name. Caches already instantiaed classes in Hashtable Cache_PostProcess. If the post process class name doesn't contain '.', then NAMESPACE_POSTPROCESS is prefixed to the name.

Parameters:

Name - The name of the post process to return.

getPreProcessModule

protected Process getPreProcessModule(java.lang.String Name)

Returns the PostProcess named Name. Caches already instantiaed classes in Hashtable Cache_PostProcess. If the post process class name doesn't contain '.', then NAMESPACE_PREPROCESS is prefixed to the name.

Parameters:

Name - The name of the post process to return.

getPostFilterModule

protected PostFilter getPostFilterModule(java.lang.String Name)

Returns the post filter class named ModelName. Caches already instantiaed matching models in Hashtable Cache_PostFilter. If the matching model name doesn't contain '.', then NAMESPACE_POSTFILTER is prefixed to the name.

Parameters:

Name - The name of the post filter to return

getInfo

public java.lang.String getInfo(SearchRequest srq)

Returns information about the weighting models and the post processors used for the given search request.

Parameters:

srq - the search request for which we obtain the information.

Returns:

String information about the weighting models and the post processors used.