Search, APIs, capability management and the Sensis journey

Embed Size (px)

Citation preview

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    1/23

    Search, APIs,Capability Managementand

    the Sensis Journey

    Craig Rees

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    2/23

    Project background Platform selection Search capability Relevance

    Architecture Quality management Hurdles Whats next

    Todays menu

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    3/23

    Sensis helps Australians find,buy and sellFrom print directories to across-platform lead generator

    Sensis publishes over 1.8Million business listings

    Two of the top 10 visited onlinesites in Australia

    (WhitePages.com.au andYellowPages.com.au)

    Sensis

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    4/23

    Business objectives

    Drive presence in the localsearch market place

    Open up the largest database ofbusiness listings in Australia

    Reduce the effort required fromlocal search developers

    Free to use, we are after thereporting

    Technology objectives

    Develop a total search platform

    Relevancy testing as part of thedevelopment lifecycle

    A framework to identify problemspaces

    Manageable platform Continuous deployments

    Project background

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    5/23

    Developer portal

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    6/23

    Platform selection

    Support for the searchcapability team

    Structured vs nonstructured data

    Deterministic vs blackbox

    Non propriety code base Community backing

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    7/23

    Unmanaged

    Adhoc

    Monitored

    Managed

    Optimized

    No resourcesNo reportingOut of the box

    features

    Adhoc processesPart time teamStatic dictionariesIndividual led innovation

    Defined teamRegular monitoringStatic autosuggestBasic linguistics

    Online dashboardsTest environmentsDynamic search refinementsTargets and metrics

    A/B testingMachine learningExternal collaborationMultiple contexts

    The Sensis Search capability maturity model*Courtesy of Pete Crawford & Craig Lonsdale

    Lvl 5

    Lvl 4

    Lvl 3

    Lvl 2

    Lvl 1

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    8/23

    Context is key

    Intent

    NameType Product Spatial

    Location

    Chronology

    Social Graph

    Individual

    Device

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    9/23

    HistoricalsearchData

    MongoDB

    BusinessData

    Geo Service

    Index

    Name QueryHandler

    Type QueryHandler

    BusinessData

    SearchService

    ReportingService

    ReportingEvents

    Publisher

    Solr

    API

    Ontologies

    Mashery

    Our architecture

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    10/23

    HistoricalsearchData

    MongoDB

    BusinessData

    Geo Service

    Index

    Name QueryHandler

    Type QueryHandler

    BusinessData

    SearchService

    ReportingService

    ReportingEvents

    Publisher

    Solr

    API

    Ontologies

    Mashery

    Data staging

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    11/23

    HistoricalsearchData

    MongoDB

    BusinessData

    Geo Service

    Index

    Name QueryHandler

    Type QueryHandler

    BusinessData

    SearchService

    ReportingService

    ReportingEvents

    Publisher

    Solr

    API

    Ontologies

    Mashery

    Search

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    12/23

    HistoricalsearchData

    MongoDB

    BusinessData

    Geo Service

    Index

    Name QueryHandler

    Type QueryHandler

    BusinessData

    SearchService

    ReportingService

    ReportingEvents

    Publisher

    Solr

    API

    Ontologies

    Mashery

    API

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    13/23

    HistoricalsearchData

    MongoDB

    BusinessData

    Geo Service

    Index

    Name QueryHandler

    Type QueryHandler

    BusinessData

    SearchService

    ReportingService

    ReportingEvents

    Publisher

    Solr

    API

    Ontologies

    Mashery

    API proxy

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    14/23

    Moved from a black box solutionto a manageable platform

    Deliver search improvementswithout major code changes

    Understand how results werecalculated

    Identity problems scientifically Continuously tune and test

    relevance

    Evolution of search management

    Yesterday Today Tomorrow

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    15/23

    Problem spaces, quality management & tuning

    Path Analysis

    used to identify

    problemsspaces

    Problem spaces, quality management & tuning

    Gold Setsused to define

    overall quality

    score (TREC)

    Features signed offonly when they make

    a positive impact to

    quality score

    Specific gold sets for eachproblem space: Intent Spelling & stemming Location Phrase parsing

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    16/23

    Search quality analysis and testing

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    17/23

    Results examiner

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    18/23

    Score analysis

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    19/23

    Tuning

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    20/23

    Lather, rinse, repeat

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    21/23

    Hurdles along the way

    Data redundancy and homogeneity

    Solr ranking of rare terms Intent differentiation Contextual synonyms

    Q i

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    22/23

    Where next?

    Query engine Facets / autosuggest Real time tuning Machine learning Multi term queries Scoring thresholds

    Content Value

  • 8/6/2019 Search, APIs, capability management and the Sensis journey

    23/23

    Questions?

    Email: [email protected]: developers.sensis.com.au

    Twitter: @SensisAPI

    @ablebagel