Exploring the Query-Flow Graph with a Mixture Model for Query Recommendation Lu Bai, Jiafeng Guo, Xueqi Cheng, Xiubo Geng, Pan Du Institute of Computing.

Exploring the Query-Flow Graph with a Mixture Model for

Query Recommendation

Lu Bai, Jiafeng Guo, Xueqi Cheng, Xiubo Geng, Pan Du

Institute of Computing Technology , CAS

Outline

• Introduction• Our approach• Experimental results• Conclusion & Future work

Introduction

• Query recommendation – Generated from web query log – Different types of information

are considered, including search results, clickthrough data, search sessions.

Introduction

• Recently, query-flow graph was introduced into query recommendation.

360 Xbox 360 kinect

360 Xbox 360 Xbox 720

Yahoo 360

Kinect Xbox 720 1 121 1

Yahoo Yahoo mail

Yahoo mail Yahoo messenger

Yahoo messenger Yahoo

1

1 1

apple Yahoo

apple apple tree

11

Introduction• Traditionally, personalized random walk over query-

flow graph was used for recommendation.• Dangling queries– No out links– Nearly 9% of whole queries

• Ambiguous queries – Mixed recommendation

• Hard to read

– Dominant recommendation• Cannot satisfy different needs

Query = 360

Xbox 360Xbox 720

Kinect

1 121 1

1

1 1

11

Query = apple

Yahoo

apple tree

Yahoo mail

Our Work

• Explore query-flow graph for better recommendation– Apply a novel mixture model over query-flow

graph to learn the intents of queries.– Perform an intent-biased random walk on the

query-flow graph for recommendation.

Probabilistic model of generating query-flow graph

• Model the generation of the query-flow graph with a novel mixture model

• Assumptions– Queries are triggered by query intents.– Consecutive queries in one search session are

from the same intent.

Probabilistic model of generating query-flow graph

• Process of generating a directed edge– Draw an intent indicator

from the multinomial distribution .

– Draw query nodes from the same multinomial intent distribution , respectively.

– Draw the directed edge from a binomial distribution

ije

ijrg

,i jq q

ije

,ij r

r

, , ,11 : ( )

Pr( | , , )ijN K

r r i r j ij rri j j C i

wG

Likelihood function

Probabilistic model of generating query-flow graph

• EM algorithm is used to estimate parameters– E step

– M step

,1 : ( )

,1 1 : ( )

N

ij ij ri j j C i

r K N

ij ij rr i j j C i

w q

w q

����������������������������

����������������������������

, ,: ( ) : ( )

,

, ,1 : ( ) : ( )

ij ij r ki ki rj j C i k i C k

r i N

ij ij r ki ki ri j j C i k i C k

w q w q

w q w q

��������������������������������������������������������

��������������������������������������������������������

,

,, ,

ij ij r

ij rij ij r ji ji r

w q

w q w q

����������������������������

����������������������������������������������������������������������

, , ,

,

, , ,1

r r i r j ij r

Kij r

r r i r j ij rr

q

��������������

��������������

��������������

Intent-biased random walk• Based on the learned query intents, we apply intent-

biased random walk for query recommendation.

– Dangling queries: back off to its intents– Ambiguous queries: recommend under the each intent

, ,(1 )i r i rA M P 1

, · (1 )Ti r i rP e ,

A row vector of query distribution of intent r

transition probability matrix

preference vector

All entries are zeroes, except that the i-th is 1

row normalized weight matrix

Experiments

• Data Set– A 3-month query log generated from a commercial

search engine.– Sessions are split by 30 minutes. – No stemming and no stop words removing.– The biggest connected graph is extracted for

experiments, which is consisted of 16,980 queries and 51,214 edges.

Experiments

• Learning performance on different intent number.

Experiments

• Learned query intents:lyrics cars poemslyrics bmw poems

song lyrics lexus love poems

lyrics com audi poetry

a z lyrics toyota friendship poems

music lyrics acura famous love poems

azlyrics nissan love quotes

lyric infiniti sad poems

az lyrics mercedes benz quotes

rap lyrics volvo mother s day poems

country lyrics mercedes mothers day poems

Experiments

• Dangling query suggestion • Ambiguous query suggestionQuery = yamaha motor

Baseline Ours

mapquest yamaha

american idol honda

yahoo mail suzuki

home depot kawasaki

bank of america yamaha motorcycles

target yamaha motorcycle

Query = hilton

Baseline Ours

marriott [hotel]

expedia marriott

holiday inn holiday inn

hyatt sheraton

hotel hampton inn

mapquest embassy suites

hampton inn hotels com

sheraton [celebrity]

hilton com paris hilton

hotels com michelle wie

embassy suites nicole richie

residence inn jessica simpson

choice hotels pamela anderson

marriot daniel dipiero

hilton honors richard hatch

Experiments

• Performance improvement based on user click behaviors

Baseline method Our approachAverage Hit Number 4.09 4.21(+2.9%)

Average Hit Score 0.598 0.652(+9.0%)

Average Score 0.181 0.194(+7.1%)

Conclusion and Future work

• conclusion– We explore the query-flow graph with a novel

probabilistic mixture model for learning query intents.

– An intent-biased random walk is introduced to integrate the learned intents for recommendation.

• Future work– Learn query intents with more auxiliary

information: clicks, URLs, words etc.