Abstract
domain models directly from natural language (NL) descriptions
of activity sequences. The modelling problem has been
identified as a bottleneck for the widespread exploitation of
various technologies in Artificial Intelligence, including automated
planners. There have been great advances in modelling
assisting and model generation tools, including a wide range
of domain model acquisition tools. However, for modelling
tools, there is the underlying assumption that the user can
formulate the problem using some formal language. And even
in the case of the domain model acquisition tools, there is
still a requirement to specify input plans in an easily machine
readable format. Providing this type of input is impractical for
many potential users. This motivates us to generate planning
domain models directly from NL descriptions, as this would
provide an important step in extending the widespread adoption
of planning techniques. We start from NL descriptions of
actions and use NL analysis to construct structured representations,
from which we construct formal representations of the
action sequences. The generated action sequences provide the
necessary structured input for inducing a PDDL domain, using
domain model acquisition technology. In order to capture
a concise planning model, we use an estimate of functional
similarity, so sentences that describe similar behaviours are
represented by the same planning operator. We validate our
approach with a user study, where participants are tasked with
describing the activities occurring in several videos. Then our
system is used to learn planning domain models using the
participants’ NL input. We demonstrate that our approach is
effective at learning models on these task
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the Twenty-Seventh International Conference on Automated Planning and Scheduling, ICAPS 2017, Pittsburgh, Pennsylvania, USA, June 18-23, 2017. |
| Publisher | AAAI press |
| Pages | 434-442 |
| Number of pages | 9 |
| Publication status | Published - 2017 |
| Externally published | Yes |
| Event | 27th International Conference on Automated Planning and Scheduling - Pittsburgh, United States Duration: 18 Jun 2017 → 23 Jun 2017 Conference number: 27 http://icaps17.icaps-conference.org/ (Link to Conference Website) |
Conference
| Conference | 27th International Conference on Automated Planning and Scheduling |
|---|---|
| Abbreviated title | ICAPS 2017 |
| Country | United States |
| City | Pittsburgh |
| Period | 18/06/17 → 23/06/17 |
| Internet address |
|
Fingerprint
Cite this
}
Framer: Planning Models from Natural Language Action Descriptions. / Lindsay, Alan; Read, Jonathon; Ferreira, João F.; Hayton, Thomas; Porteous, Julie; Gregory, Peter.
Proceedings of the Twenty-Seventh International Conference on Automated Planning and Scheduling, ICAPS 2017, Pittsburgh, Pennsylvania, USA, June 18-23, 2017.. AAAI press, 2017. p. 434-442.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Framer: Planning Models from Natural Language Action Descriptions
AU - Lindsay, Alan
AU - Read, Jonathon
AU - Ferreira, João F.
AU - Hayton, Thomas
AU - Porteous, Julie
AU - Gregory, Peter
PY - 2017
Y1 - 2017
N2 - In this paper, we describe an approach for learning planningdomain models directly from natural language (NL) descriptionsof activity sequences. The modelling problem has beenidentified as a bottleneck for the widespread exploitation ofvarious technologies in Artificial Intelligence, including automatedplanners. There have been great advances in modellingassisting and model generation tools, including a wide rangeof domain model acquisition tools. However, for modellingtools, there is the underlying assumption that the user canformulate the problem using some formal language. And evenin the case of the domain model acquisition tools, there isstill a requirement to specify input plans in an easily machinereadable format. Providing this type of input is impractical formany potential users. This motivates us to generate planningdomain models directly from NL descriptions, as this wouldprovide an important step in extending the widespread adoptionof planning techniques. We start from NL descriptions ofactions and use NL analysis to construct structured representations,from which we construct formal representations of theaction sequences. The generated action sequences provide thenecessary structured input for inducing a PDDL domain, usingdomain model acquisition technology. In order to capturea concise planning model, we use an estimate of functionalsimilarity, so sentences that describe similar behaviours arerepresented by the same planning operator. We validate ourapproach with a user study, where participants are tasked withdescribing the activities occurring in several videos. Then oursystem is used to learn planning domain models using theparticipants’ NL input. We demonstrate that our approach iseffective at learning models on these task
AB - In this paper, we describe an approach for learning planningdomain models directly from natural language (NL) descriptionsof activity sequences. The modelling problem has beenidentified as a bottleneck for the widespread exploitation ofvarious technologies in Artificial Intelligence, including automatedplanners. There have been great advances in modellingassisting and model generation tools, including a wide rangeof domain model acquisition tools. However, for modellingtools, there is the underlying assumption that the user canformulate the problem using some formal language. And evenin the case of the domain model acquisition tools, there isstill a requirement to specify input plans in an easily machinereadable format. Providing this type of input is impractical formany potential users. This motivates us to generate planningdomain models directly from NL descriptions, as this wouldprovide an important step in extending the widespread adoptionof planning techniques. We start from NL descriptions ofactions and use NL analysis to construct structured representations,from which we construct formal representations of theaction sequences. The generated action sequences provide thenecessary structured input for inducing a PDDL domain, usingdomain model acquisition technology. In order to capturea concise planning model, we use an estimate of functionalsimilarity, so sentences that describe similar behaviours arerepresented by the same planning operator. We validate ourapproach with a user study, where participants are tasked withdescribing the activities occurring in several videos. Then oursystem is used to learn planning domain models using theparticipants’ NL input. We demonstrate that our approach iseffective at learning models on these task
M3 - Conference contribution
SP - 434
EP - 442
BT - Proceedings of the Twenty-Seventh International Conference on Automated Planning and Scheduling, ICAPS 2017, Pittsburgh, Pennsylvania, USA, June 18-23, 2017.
PB - AAAI press
ER -