Managing General and Individual Knowledge in Crowd Mining Applications Yael Amsterdamer, Susan Davidson, Anna Kukliansky, Tova Milo, Slava Novgorodov and Amit Somech CIDR 2015
Managing General and Individual Knowledge in Crowd Mining Applications 2 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice).
Managing General and Individual Knowledge in Crowd Mining Applications 3 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. Tips: Apply for a ballfield permit online” “You can go visit the Bronx Zoo and eat at Pine Restaurant. Tips: Order antipasti at Pine. Skip dessert and go for ice cream across the street”
Managing General and Individual Knowledge in Crowd Mining Applications 4 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. Tips: Apply for a ballfield permit online” “You can go visit the Bronx Zoo and eat at Pine Restaurant. Tips: Order antipasti at Pine. Skip dessert and go for ice cream across the street”
Managing General and Individual Knowledge in Crowd Mining Applications 5 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. Tips: Apply for a ballfield permit online” “You can go visit the Bronx Zoo and eat at Pine Restaurant. Tips: Order antipasti at Pine. Skip dessert and go for ice cream across the street”
Managing General and Individual Knowledge in Crowd Mining Applications 6 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. Tips: Apply for a ballfield permit online” “You can go visit the Bronx Zoo and eat at Pine Restaurant. Tips: Order antipasti at Pine. Skip dessert and go for ice cream across the street” A dietician may wish to study the culinary preferences in some population, focusing on food dishes that are rich in fiber
Managing General and Individual Knowledge in Crowd Mining Applications 7 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. Tips: Apply for a ballfield permit online” “You can go visit the Bronx Zoo and eat at Pine Restaurant. Tips: Order antipasti at Pine. Skip dessert and go for ice cream across the street” A dietician may wish to study the culinary preferences in some population, focusing on food dishes that are rich in fiber
Managing General and Individual Knowledge in Crowd Mining Applications 8 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. General knowledge: Individual knowledge: Tips: Apply for a ballfield permit online” • General truth, objective data, not • Related to the habits and opinions “You can go visit the Bronx Zoo and eat at Pine Restaurant. associated with an individual of an individual • E.g., geographical locations Tips: Order antipasti at Pine. • E.g., travel recommendations • Can be found in a knowledge base Skip dessert and go for ice cream across the street” • We can ask people about it or an ontology A dietician may wish to study the culinary preferences in some population, focusing on food dishes that are rich in fiber
Managing General and Individual Knowledge in Crowd Mining Applications 9 Motivation Ann, a vacationer, is interested in finding child-friendly activities at an attraction in NYC, and a good restaurant nearby (plus relevant advice). “You can play baseball in Central Park and eat at Maoz Vegetarian. General knowledge: Individual knowledge: Tips: Apply for a ballfield permit online” • General truth, objective data, not • Related to the habits and opinions “ You can go visit the Bronx Zoo and eat at Pine Restaurant. associated with an individual of an individual • E.g., geographical locations Tips: Order antipasti at Pine. • E.g., travel recommendations • Can be found in a knowledge base Skip dessert and go for ice cream across the street” • We can ask people about it or an ontology When missing in the knowledge base, Crowd answers can be recoded in a A dietician may wish to study the culinary preferences in some we can ask the crowd! knowledge base population, focusing on food dishes that are rich in fiber
Managing General and Individual Knowledge in Crowd Mining Applications 10 Crowd Mining: Crowdsourcing in an Open World Given an ontology of general knowledge and a mining task • Incrementally explore relevant patterns {Ball_Game playAt Central_Park} • Generate (closed and open) questions to the crowd about them How often do you play ball games Which ball games do you play at Central Park ? at Central Park ? What else do you do at Central Park ? • Evaluate the significance of the patterns and discover related ones Pattern score = 0.6 {Baseball playAt Central_Park. Permit getAt "www.permits.org"} • Produce a concise output that summarizes the findings
Managing General and Individual Knowledge in Crowd Mining Applications 11 Crowd Mining Framework Design We design a general architecture which outlines the components of a crowd mining framework and the interaction between them Challenges: Compiling user requests into Deciding which questions to a declarative query language generate to the crowd next How to aggregate Personalization and Updating and managing crowd answers? crowd member selection the knowledge base The type of processed data Combining the crowd answers (general versus individual) with knowledge base data must be taken into account
Managing General and Individual Knowledge in Crowd Mining Applications 12 Today Motivation Framework Architecture Zoom-in on components Examples via the OASSIS system
Managing General and Individual Knowledge in Crowd Mining Applications 13 The Architecture Knowledge Base budget, preferences Query Knowledge Input general query Engine Inferred general NL request query updates user NL Parser / Generator Inferred Tasks, budget, Significant Request individual preferences results User Interface refinement knowledge Crowd Task updates Crowd results Summarized Manager crowd results Result Summarized Answer summary general aggregation Summarized Inference and NL task task individual summarization Significance Per worker function NL answer result updates Raw crowd User data task Overall results Utility Raw crowd result results Crowd Task, workers Next Crowd preferences worker Input Crowd Selection Inferred Crowd worker reward properties User/worker Profile
Managing General and Individual Knowledge in Crowd Mining Applications 14 Knowledge Repository Different types of knowledge: • A general knowledge base is input to the system • Knowledge inferred in previous query evaluation Input general – General knowledge – completes the knowledge base Inferred general May be annotated with trust/error probability Inferred – Individual knowledge – more volatile individual may be annotated with user properties
Managing General and Individual Knowledge in Crowd Mining Applications 15 Knowledge Repository Different types of knowledge: • A general knowledge base is input to the system • Knowledge inferred in previous query evaluation can be recorded – General knowledge – completes the knowledge base May be annotated with trust/error probability – Individual knowledge – more volatile may be annotated with user properties
Managing General and Individual Knowledge in Crowd Mining Applications 16 Knowledge Repository Different types of knowledge: • A general knowledge base is input to the system Input general Inferred • Knowledge inferred in previous query evaluation Grimaldi's general can be recorded Inferred individual – General knowledge – completes the knowledge base Shake May be annotated with trust/error probability Shack – Individual knowledge – more volatile may be annotated with user properties nearby
Managing General and Individual Knowledge in Crowd Mining Applications 17 Knowledge Repository Different types of knowledge: • A general knowledge base is input to the system People Frequently eat at Input general Inferred • Knowledge inferred in previous query evaluation Grimaldi's general can be recorded Inferred individual – General knowledge – completes the knowledge base Shake May be annotated with trust/error probability Shack – Individual knowledge – more volatile may be annotated with user properties nearby
Recommend
More recommend
