Package com.yahoo.prelude.query

Interface TaggableItem

All Known Implementing Classes:

BoolItem, CompositeIndexedItem, CompositeTaggableItem, DotProductItem, EquivItem, ExactStringItem, IndexedSegmentItem, IntItem, MarkerWordItem, NearestNeighborItem, PhraseItem, PhraseSegmentItem, PredicateQueryItem, PrefixItem, RangeItem, RegExpItem, SimpleIndexedItem, SimpleTaggableItem, SubstringItem, SuffixItem, TaggableSegmentItem, TermItem, UriItem, WandItem, WeightedSetItem, WordAlternativesItem, WordItem

public interface TaggableItem

An interface used for anything which may be addressed using an external, unique ID in the query tree in the backend.

Author:

Steinar Knutsen

Method Summary

Modifier and Type	Method	Description
Item	getConnectedItem()
double	getConnectivity()
double	getSignificance()
int	getUniqueID()
boolean	hasExplicitSignificance()
boolean	hasUniqueID()
void	setConnectivity(Item item, double connectivity)	Set the connectivity to another term in the same query tree.
void	setExplicitSignificance(boolean significance)
void	setSignificance(double significance)	Used for setting explicit term significance (in the tf/idf sense) to a single term or phrase, relative to the rest of the query.
void	setUniqueID(int id)

Method Detail

getUniqueID

int getUniqueID()

setUniqueID

void setUniqueID(int id)

hasUniqueID

boolean hasUniqueID()

setConnectivity

void setConnectivity(Item item,
                     double connectivity)

Set the connectivity to another term in the same query tree. This is used to influence ranking features taking proximity into account: nativeRank and a subset of the fieldMatch features.

By default consecutive query terms are 'somewhat' connected, meaning ranking features will be better in documents where the terms are found close to each other. This effect can be increased or decreased by manipulating the connectivity value. Typical use is to increase the connectivity between terms in the query that we believe are semantically connected. E.g in the query 'new york hotel', it is a good idea to increase the connectivity between "new" and "york" to ensure that a document containing "List of hotels in New York" is ranked above one containing "List of new hotels in York".

Parameters:

item - the item this should be connected to - in practice the next consecutive item in the query

connectivity - a value between 0 (none) and 1 (maximal), defining the connectivity between this and the argument item. The default connectivity is 0.1.

getConnectedItem

Item getConnectedItem()

getConnectivity

double getConnectivity()

setSignificance

void setSignificance(double significance)

Used for setting explicit term significance (in the tf/idf sense) to a single term or phrase, relative to the rest of the query. This influences ranking features which take term significance into account and overrides the default partial corpus based term significance computation happening in the backend.

hasExplicitSignificance

boolean hasExplicitSignificance()

setExplicitSignificance

void setExplicitSignificance(boolean significance)

getSignificance

double getSignificance()