Does a shared screen make a shared understanding

  • Slides: 22
Download presentation
Does a shared screen make a shared understanding ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs.

Does a shared screen make a shared understanding ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

NO Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

NO Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

IS W I S WY shared screen shared understandinging d n u o r

IS W I S WY shared screen shared understandinging d n u o r G shared understanding e v i t a r bo Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu Colla ing learn

WHO KILLED MONA-LISA? Restaurant Room 5 Note Kitchen Colonel Von Lucie Salève Schneider ylster

WHO KILLED MONA-LISA? Restaurant Room 5 Note Kitchen Colonel Von Lucie Salève Schneider ylster Gun O Lobby Hans Wenger Heidi Zeller Marie Salève Registr ce Insuran Entrance y et k Ski jac g Paintin Giuzeppe Vesuvio Room 8 og Private Jacques Salève Room 7 Phone. L Oscar Salève Bar Room 6 Lisa Jones Claire & Rolf Loretan Room 1 Room 2 Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu Room 3 MLV Room 4 Gun

MOO Auberge Guest Room: 4 You enter a largue blue room with a small

MOO Auberge Guest Room: 4 You enter a largue blue room with a small window. You see Helmut, Ski Jacket, and Gun here. Obvious Exits: Out (to Lower Corridor). > look gun You see an old Swiss army pistol. > ask helmut about last night I stayed at the bar until 9 Pm and then went to bed. > page sherlock Interesting isn't? Sherlock has received your page. He pages « He lies » . Sherlock joins you. Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

MOO Whiteboard > " Hi colleague > ' Where are you? > ask MS

MOO Whiteboard > " Hi colleague > ' Where are you? > ask MS about last night > look gun > ask MS about last gun > read insurance > read all from DN 2 > read Hans from DN 2 > compare DN 1 with DN 2 MOO > " skjhkjh dfsdfsf > ask Helmut about last night > ask MS about mona > look painting > read all from DN 1 > read Hans from DN 1 > compare DN 1 with DN 1 20 pairs Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu Whiteboard

? How does the whiteboard help to • ground utterances • share solutions Pierre.

? How does the whiteboard help to • ground utterances • share solutions Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Whiteboard roles: • envisionning solutions • grouding utterances MOO > " skjhkjh dfsdfsf •

Whiteboard roles: • envisionning solutions • grouding utterances MOO > " skjhkjh dfsdfsf • explanatory schema > ask Helmut about last night > ask MS about mona > look painting • deictics • shared memory • regulation Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu > read all from DN 1 > read Hans from DN 1 > compare DN 1 with DN 1 Whiteboard

The whiteboard is the central space for sharing information (in this task) Pierre. Dillenbourg@tecfa.

The whiteboard is the central space for sharing information (in this task) Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Knowledge Whiteboard Display Persistent Non-Persistent Non Persistent MOO dialogues Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs.

Knowledge Whiteboard Display Persistent Non-Persistent Non Persistent MOO dialogues Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Grounding varies according to problem solving variables Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Grounding varies according to problem solving variables Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Acknowlegment in MOO dialogues • Rate = 41% (90% in voice experiments) • Not

Acknowlegment in MOO dialogues • Rate = 41% (90% in voice experiments) • Not related with MOO expertise • Not related with interactions parameters • symmetry(ack+): 8% = • ack (short delay): 41% = • ack (freq+. talkers): 41% = • turns. IC (ack+): 0. 9 = symmetry(ack-): 8% ack (long delay): 41%; ack (freq-. talkers): 42%; turns. IC (ack+): 0. 9 • Related with problem sloving • N. actions (ack+): 178 <<. 05 N. actions (ack-) : 237 • redudancy (ack+): 6 <<. 01 redudancy (ack-): 18 Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

Degree of sharedness Visibility Understanding Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu Facts in

Degree of sharedness Visibility Understanding Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu Facts in whiteboard Inferences in whiteboard Acces Inferences in talk Facts in talk Agreement Rate of acknowledgment

. . and the subjects: • maintain multiple conversational contexts • are sensitive to

. . and the subjects: • maintain multiple conversational contexts • are sensitive to space beyond any functional constraint • negotiate across modalities • (re-)allocate functions to tools in various ways Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

MOO dialogue MOO action Collect facts Share inferences Store facts Store inferences Coordinate action

MOO dialogue MOO action Collect facts Share inferences Store facts Store inferences Coordinate action Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu White board Note book

Distributed Cognition Reconfigurable because distributed Shared despite distrbuted Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd.

Distributed Cognition Reconfigurable because distributed Shared despite distrbuted Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

How do agents coconstruct culture ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

How do agents coconstruct culture ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

WYSIWIS-relaxed + awareness tools jhk slk hlkjsh lkjhsdl kgjhnbjhb sh rgl hc , cg

WYSIWIS-relaxed + awareness tools jhk slk hlkjsh lkjhsdl kgjhnbjhb sh rgl hc , cg lkw kjehéeéhlk h jhk slk hlkjsh lkjhsdl Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

shared understanding ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu

shared understanding ? Pierre. Dillenbourg@tecfa. unige. ch, Traum@cs. umd. edu