From arunbkris at gmail.com Fri Jul 10 07:25:17 2020 From: arunbkris at gmail.com (Arun Krishna) Date: Fri, 10 Jul 2020 13:25:17 +0200 Subject: [ACT-R-users] Support needed for using ACT-R RPC Message-ID: Hi, My java application needs to communicate to ACT-R cognition models via RPC. In a big picture level Java application -> Sent the current context information -> ACT-R run the cognition models ->Get back the result from ACT-R . I had gone through http://act-r.psy.cmu.edu/actr7.x/remote.pdf as well as some examples given in \ACT-R\examples\connections\lisp . But I didn't understand enough to use the RPC. The current examples contain only two method invocations print("{\"method\":\"set-name\",\"params\":[\"Simple Java Example\"],\"id\":null}"); // Evaluate the ACT-R "act-r-version" command. print("{\"method\":\"evaluate\",\"params\":[\"act-r-version\"],\"id\":1}"); Is there an example which does more han this with RPC, i.e communication to a count model etc? Thanks & Regards, Arun -------------- next part -------------- An HTML attachment was scrubbed... URL: From db30 at andrew.cmu.edu Fri Jul 10 10:37:51 2020 From: db30 at andrew.cmu.edu (Dan Bothell) Date: Fri, 10 Jul 2020 10:37:51 -0400 Subject: [ACT-R-users] Support needed for using ACT-R RPC In-Reply-To: References: Message-ID: <1a41d924-50cc-e5d2-fd8c-a4f6ac3de12d@andrew.cmu.edu> On 7/10/20 7:25 AM, Arun Krishna wrote: > Hi, > My java application needs to communicate to ACT-R cognition models via > RPC. > In a big picture level > Java application -> Sent the current context information -> ACT-R run > the cognition models ->Get back the result from ACT-R . > I had gone through http://act-r.psy.cmu.edu/actr7.x/remote.pdf as well > as some examples given in > \ACT-R\examples\connections\lisp . But I didn't understand enough to > use the RPC. > The current examples contain only two method invocations > print("{\"method\":\"set-name\",\"params\":[\"Simple Java > Example\"],\"id\":null}"); > > // Evaluate the ACT-R "act-r-version" command. print("{\"method\":\"evaluate\",\"params\":[\"act-r-version\"],\"id\":1}"); > > Is there an example which does more han this with RPC, i.e communication to a count model etc? > > Thanks & Regards, > Arun Given what you describe, it would seem the evaluate method and the corresponding result message sent back may be all you need. The evaluate method is used to call commands that are available through the RPC system, and the return value(s) are sent back using result. Thus, you could evaluate ACT-R commands to set the context (add chunks, set buffer contents, create visual features, etc), evaluate one of the running commands to run the model, and then evaluate commands to get information from the model, like the the contents of buffers or the current model time. The ACT-R reference manual contains details of all the commands available remotely.? Also, if you haven't done so, I would recommend working through the ACT-R tutorial to see practical examples of how many of the commands are used to setup and run models.? Included with the tutorial is also an implementation of the ACT-R RPC protocol in Python which provides Python functions to call specific ACT-R commands, which may be helpful in understanding how to use it. Hope that helps, and let me know if you have further questions, Dan -------------- next part -------------- An HTML attachment was scrubbed... URL: From arunbkris at gmail.com Sat Jul 11 13:58:16 2020 From: arunbkris at gmail.com (Arun Krishna) Date: Sat, 11 Jul 2020 19:58:16 +0200 Subject: [ACT-R-users] Support needed for using ACT-R RPC In-Reply-To: <1a41d924-50cc-e5d2-fd8c-a4f6ac3de12d@andrew.cmu.edu> References: <1a41d924-50cc-e5d2-fd8c-a4f6ac3de12d@andrew.cmu.edu> Message-ID: Hi Dan, Thank you very much for the reply. Now I understand that the evaluate method serves the purpose. Can you please give me a simple example invoking add chunk using evaluate if it is not taking too much of your time? The examples you are referring is ACT-R\examples\connections\python I assume, but this one gives a reference for invoking act-r-version "send(sock,'{"method":"evaluate","params":["act-r-version"],"id":1}')" Thanks & Regards, Arun On Fri, Jul 10, 2020 at 4:39 PM Dan Bothell wrote: > > On 7/10/20 7:25 AM, Arun Krishna wrote: > > Hi, > My java application needs to communicate to ACT-R cognition models via RPC. > In a big picture level > Java application -> Sent the current context information -> ACT-R run the > cognition models ->Get back the result from ACT-R . > I had gone through http://act-r.psy.cmu.edu/actr7.x/remote.pdf as well as > some examples given in > \ACT-R\examples\connections\lisp . But I didn't understand enough to use > the RPC. > The current examples contain only two method invocations > > print("{\"method\":\"set-name\",\"params\":[\"Simple Java Example\"],\"id\":null}"); > // Evaluate the ACT-R "act-r-version" command.print("{\"method\":\"evaluate\",\"params\":[\"act-r-version\"],\"id\":1}"); > > > Is there an example which does more han this with RPC, i.e communication to a count model etc? > > > Thanks & Regards, > > Arun > > > Given what you describe, it would seem the evaluate method and the > corresponding result message sent back may be all you need. The evaluate > method is used to call commands that are available through the RPC system, > and the return value(s) are sent back using result. > > Thus, you could evaluate ACT-R commands to set the context (add chunks, > set buffer contents, create visual features, etc), evaluate one of the > running commands to run the model, and then evaluate commands to get > information from the model, like the the contents of buffers or the current > model time. > > The ACT-R reference manual contains details of all the commands available > remotely. Also, if you haven't done so, I would recommend working through > the ACT-R tutorial to see practical examples of how many of the commands > are used to setup and run models. Included with the tutorial is also an > implementation of the ACT-R RPC protocol in Python which provides Python > functions to call specific ACT-R commands, which may be helpful in > understanding how to use it. > > Hope that helps, and let me know if you have further questions, > Dan > _______________________________________________ > ACT-R-users mailing list > ACT-R-users at act-r.psy.cmu.edu > https://mailman.srv.cs.cmu.edu/mailman/listinfo/act-r-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From arunbkris at gmail.com Mon Jul 13 16:49:25 2020 From: arunbkris at gmail.com (Arun Krishna) Date: Mon, 13 Jul 2020 22:49:25 +0200 Subject: [ACT-R-users] Support needed for using ACT-R RPC In-Reply-To: References: <1a41d924-50cc-e5d2-fd8c-a4f6ac3de12d@andrew.cmu.edu> Message-ID: Hi Dan, Thank you very much for the guidance as well as pointing me to the correct example . The examples in tutoral/python is very helpful. Regards, Arun On Mon, Jul 13, 2020 at 5:14 PM wrote: > > > --On Saturday, July 11, 2020 1:58 PM -0400 Arun Krishna < > arunbkris at gmail.com> > wrote: > > > > > > > Hi Dan, > > > > > > Thank you very much for the reply. > > Now I understand that the evaluate method serves the purpose. > > Can you please give me a simple example invoking add chunk using > evaluate if > > it is not taking too much of your time? > > The examples you are referring is ACT-R\examples\connections\python I > assume, > > but this one gives a reference for invoking act-r-version > > > > "send(sock,'{"method":"evaluate","params":["act-r-version"],"id":1}')" > > > > > > > The Python interface provided with the tutorial is in the tutoral/python > directory, and the file actr.py implements the interface. > > There isn't a command called "add chunk". There are commands called > "define-chunks", "add-dm", "add-dm-fct", and possibly others depending > on exactly what you want to happen. You can find the details in the > reference manual along with the parameters that they would be passed. > > To evaluate one you would send the the evaluate method with the params > set to a list with the name of the command, the name of a model or a > null/false marker to indicate the current model, then the parameters > as needed for the command. > > Dan > -------------- next part -------------- An HTML attachment was scrubbed... URL: From coty at cmu.edu Tue Jul 14 12:56:48 2020 From: coty at cmu.edu (Cleotilde Gonzalez) Date: Tue, 14 Jul 2020 16:56:48 +0000 Subject: [ACT-R-users] New Post-Doctoral Fellow Position at Carnegie Mellon University Message-ID: <17e929c11c494010b9ca614f479e31b4@cmu.edu> Dear Colleagues, My lab (www.cmu.edu/ddmlab) has a post-doctoral position slot available to start any time between August 17, 2020 to September 3, 2020. The Post-doctoral fellow will work on the general area of Human-Machine Teaming. He/She will support the design of experimental studies involving one or more (networked) participants; and will also need to be engaged in the development of computational models of individual and collective behavior. Required qualifications: -- A Ph.D. (completed by start of employment) in cognitive science or behavioral psychology, human factors engineering, or equivalent. Technical background is required; for example, a Ph.D. in social psychology with no other degree in Engineering or Computer Science would not be acceptable. -- Training in behavioral science research methods and statistical analyses. -- Experience with statistical software (preferably R, others acceptable). -- Experience with computational/cognitive modeling (e.g., reinforcement learning, ACT-R models, IBL models). -- Demonstrable writing abilities through publications and good communication skills. Desired qualifications: -- Experience with programming (preferably Python, others acceptable) -- Experience with web programming and design -- Experience working with interdisciplinary, collaborative teams, and managing research assistants. Duration: This is a full time research position with full benefits, for one year with a possibility of renewal for one more additional year conditional on performance and availability of funds. Please submit: 1. A letter of interest 2. Your updated CV 3. 3 letters of recommendation 4. One research paper Directly to: coty at cmu.edu before July 31, 2020. Prof. Cleotilde Gonzalez Research Professor of Decision Science Founding Director of Dynamic Decision Making Laboratory Social and Decision Sciences Department Carnegie Mellon University 5000 Forbes Ave. Porter Hall 208 Pittsburgh, PA, 15213 phone: 412-268-6242 Fax: 412-268-6938 E-mail: coty at cmu.edu www.cmu.edu/ddmlab -------------- next part -------------- An HTML attachment was scrubbed... URL: From greg.trafton at nrl.navy.mil Tue Jul 14 14:47:28 2020 From: greg.trafton at nrl.navy.mil (Greg Trafton) Date: Tue, 14 Jul 2020 14:47:28 -0400 Subject: [ACT-R-users] Postdocs available at NRL Message-ID: <892C65EF-2BB8-4C8F-8A40-2AAE82551C28@nrl.navy.mil> Interested in doing a postdoc at an interesting, stable, and safe location? We are looking for postdocs even during the pandemic (and we offer health care). The Navy Center for Applied Research in Artificial Intelligence (NCARAI) at the Naval Research Laboratory (NRL) in Washington DC is looking for 1-2 postdocs to join our team. Our team is multidisciplinary and includes cognitive scientists, roboticists, and computer scientists. Our lab's focus is primarily on human robot interaction from a computational cognitive and social perspective. We have openings for postdocs who specialize in: * Cognitive Modeling * Human Robot Interaction * Cognitive Robotics * Agency * Social norms and social modeling * Teaming and teamwork * Reasoning * Embodied cognition We are currently recruiting for projects that focus on interactive perceptual learning, cognitive/social interaction, and robot teaming. We will consider people with expertise in any area of cognitive science, cognitive modeling, social sciences, or robotics, though an open mind, an ability to work well in medium size groups, and a strong publishing record are all required. Our postdoc deadline is August 1, so please contact me immediately (greg.trafton at nrl.navy.mil) if you are interested in becoming a postdoc with us. Our postdoc positions typically last for 2-3 years. You must be a US citizen to apply for this position. Pay is ~80K and includes health insurance, a moving allowance, and a travel allowance; details about the postdoc program itself are available at https://www.nrl.navy.mil/careers/post-doc/nrc We have excellent robotic resources and labs for interesting, innovative research with robotics. Do you have a cognitive model that needs a real body? We have a great opportunity for you. We also are working on combining cognitive models with AI models (especially deep networks) focusing on predicting human performance, individual differences, and learning. For more details about me and our research: http://gregtrafton.com/about/ http://www.nrl.navy.mil/itd/aic/content/cognitive-robotics-and-human-robot-interaction For more details on our facility: http://www.nrl.navy.mil/lasr/ For information about postdocs: http://gregtrafton.com/postdocs/ Looking forward to hearing from you, Greg Trafton, Ph.D greg.trafton at nrl.navy.mil From cl at cmu.edu Mon Jul 20 15:34:37 2020 From: cl at cmu.edu (Christian Lebiere) Date: Mon, 20 Jul 2020 15:34:37 -0400 Subject: [ACT-R-users] 2020 ACT-R Workshop In-Reply-To: References: Message-ID: A final schedule for this Saturday's ACT-R Workshop is attached and can also be found on the ACT-R website . Please let me know if you spot any issues or errors. Attending the workshop requires registering for the MathPsych/ICCM conference. Today is the last day to register at the Standard Fee. If you have registered, you should have received instructions on how to connect. The ACT-R Workshop will take place in the Toronto room that can be accessed from the virtual conference site . We will start gathering at 10:30am ET to provide ample time to be admitted to the Zoom room before the 11am start. You should download and test the Zoom app in advance if you are not a regular Zoom user. The basic rules of the workshop are as follows: 1. to prevent interference and preserve bandwidth, everyone other than speakers/panelists should be on mute with the camera off 2. attendees should post questions and comments in the chat window during the talks and panel discussions 3. the session chair will pick questions from the chat and call on people who can then unmute (and turn on camera if desired) to ask their question Each talk session will feature 3 presentations lasting up to 20' each (including questions) followed by a 20' panel discussion on the theme of the session. Breaks between sessions will last 20'. The concluding Future of ACT-R session will feature a software update by Dan Bothell followed by an open discussion. Let me know if you have any questions or encounter any issue. Hope to see you on Saturday, Christian On Mon, Jun 29, 2020 at 12:13 PM Christian Lebiere wrote: > A draft of the 2020 ACT-R Workshop schedule is attached. The workshop will > take place online on July 25, as originally planned, roughly between the > hours of 11am and 5pm ET to accommodate a range of time zones. As usual, we > would like to limit presentations to 15-20 minutes to leave ample room for > discussions. Presentations and discussions will be live, but we should be > able to accommodate pre-recorded talks if needed to address specific > concerns (e.g., connection reliability). Additional details such as > connection instructions will be forthcoming shortly. For now, please check > the draft schedule and let me know: > > - if you see your name listed correctly, please send me the title of your > talk if you haven't yet > - if you had requested a talk and do not see your name, please accept my > apologies and contact me again > - if you had not requested a talk but would like to present, please > contact me asap with the title and topic of your talk > > Finally, the early bird registration deadline for the MathPsych/ICCM > conference is > July 1. > > Best, > Christian > > On Thu, May 21, 2020 at 12:31 PM Christian Lebiere wrote: > >> Following positive feedback to our recent survey, there will be a virtual >> ACT-R Workshop this year associated with the MathPsych/ICCM conference. >> Interested participants should send presentation titles or suggestions for >> session topics to me at cl at cmu.edu. Proceedings of past workshops can be >> found at http://act-r.psy.cmu.edu/workshops/. Technical details of the >> virtual workshop are still being discussed. Comments and suggestions in >> that respect are welcome as well. >> >> Best, >> Christian >> > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: ACT-R Workshop 2020 Schedule.pdf Type: application/pdf Size: 100481 bytes Desc: not available URL: From arunbkris at gmail.com Sun Jul 26 13:24:43 2020 From: arunbkris at gmail.com (Arun Krishna) Date: Sun, 26 Jul 2020 19:24:43 +0200 Subject: [ACT-R-users] Question regarding buffer-chunk command via RPC Message-ID: Hi, My java application uses RPC to communicate with ACT-R. Mainly uses the evaluate method and is working in many cases. But buffer-chunk command is not working via evaluate eventhough buffer-read command works. For example ----------------------------------------------- JSONArray jsonArray3 = new JSONArray(); jsonArray3.put("{\"method\":\"evaluate\","); jsonArray3.put("\"params\":"); jsonArray3.put("[\"buffer-read\",\"maneuver-left\",\"goal\"],"); jsonArray3.put("\"id\":1}"); Works and I am getting proper out put {"result": ["D-0"], "error": null, "id": 1} -------------------------------------------------- ------------------------------------------------------------ But when I use buffer-chunk JSONArray jsonArray3 = new JSONArray(); jsonArray3.put("{\"method\":\"evaluate\","); jsonArray3.put("\"params\":"); jsonArray3.put("[\"buffer-chunk\",\"maneuver-left\",\"goal\"],"); jsonArray3.put("\"id\":1}"); I am getting {"result": [["D-0"]], "error": null, "id": 1} which does not include the contents in the goal buffer. When I use the same command via ACT-R command line, buffer-chunk works and gives complete output (buffer-chunk goal) GOAL: D-0 D-0 STATE3 "active" (D-0) ------------------------------------------------------------------ Can you please let me know how to get the full contents in the goal/retrieval buffer via RPC? (Note: I am using ACT-R\examples\connections\Java\simpleACTRtest.java for testing) Thanks & Regards, Arun -------------- next part -------------- An HTML attachment was scrubbed... URL: From db30 at andrew.cmu.edu Sun Jul 26 21:36:20 2020 From: db30 at andrew.cmu.edu (Dan Bothell) Date: Sun, 26 Jul 2020 21:36:20 -0400 Subject: [ACT-R-users] Question regarding buffer-chunk command via RPC In-Reply-To: References: Message-ID: <157d84af-3b68-afbd-d777-82d57fe14b31@andrew.cmu.edu> That is the correct return value from buffer-chunk when given a buffer name.? It returns a list of the names of the chunks in the buffers provided, and that is all that is returned in Lisp as well -- the (D-0) at the end.? The output to the command-trace with the buffer's chunk information is a side effect of calling buffer-chunk, and you will see that when you call it remotely as well. If you want the text output that is shown, you can use the printed-buffer-chunk command.? If you want to know what slots a chunk has, then you would use the chunk-filled-slots-list command, and to get the value of a slot in the chunk you would use the chunk-slot-value command. Dan On 7/26/20 1:24 PM, Arun Krishna wrote: > Hi, > My java application uses RPC to communicate with ACT-R. > Mainly uses the evaluate method and is working in many cases. > But buffer-chunk command is not working via evaluate eventhough > ?buffer-read command works. > > For example > ----------------------------------------------- > JSONArray jsonArray3 = new JSONArray(); > jsonArray3.put("{\"method\":\"evaluate\","); > jsonArray3.put("\"params\":"); > jsonArray3.put("[\"buffer-read\",\"maneuver-left\",\"goal\"],"); > jsonArray3.put("\"id\":1}"); > > Works and I am getting proper out put > {"result": ["D-0"], "error": null, "id": 1} > -------------------------------------------------- > > > ------------------------------------------------------------ > But when I use buffer-chunk > ?JSONArray jsonArray3 = new JSONArray(); > ? ? jsonArray3.put("{\"method\":\"evaluate\","); > ? ? jsonArray3.put("\"params\":"); > jsonArray3.put("[\"buffer-chunk\",\"maneuver-left\",\"goal\"],"); > ? ? jsonArray3.put("\"id\":1}"); > > I am getting {"result": [["D-0"]], "error": null, "id": 1} which does > not include the contents in the > goal buffer. > > When I use the same command via ACT-R command line, buffer-chunk works > and gives complete output > (buffer-chunk goal) > GOAL: D-0 > D-0 > ? ?STATE3 ?"active" > > (D-0) > ------------------------------------------------------------------ > > Can you please let me know how to get the full contents in the > goal/retrieval buffer via RPC? > (Note: I am using ACT-R\examples\connections\Java\simpleACTRtest.java > for testing) > > Thanks & Regards, > Arun > > > > _______________________________________________ > ACT-R-users mailing list > ACT-R-users at act-r.psy.cmu.edu > https://mailman.srv.cs.cmu.edu/mailman/listinfo/act-r-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From arunbkris at gmail.com Tue Jul 28 02:25:55 2020 From: arunbkris at gmail.com (Arun Krishna) Date: Tue, 28 Jul 2020 08:25:55 +0200 Subject: [ACT-R-users] Question regarding buffer-chunk command via RPC In-Reply-To: <157d84af-3b68-afbd-d777-82d57fe14b31@andrew.cmu.edu> References: <157d84af-3b68-afbd-d777-82d57fe14b31@andrew.cmu.edu> Message-ID: Hi Dan, Thank you very much. I am able to get the filled slots using chunk-filled-slots-list command and get the values of the slot successfully by using chunk-slot-value command. Regards, Arun On Mon, Jul 27, 2020 at 3:36 AM Dan Bothell wrote: > That is the correct return value from buffer-chunk when given a buffer > name. It returns a list of the names of the chunks in the buffers > provided, and that is all that is returned in Lisp as well -- the (D-0) at > the end. The output to the command-trace with the buffer's chunk > information is a side effect of calling buffer-chunk, and you will see that > when you call it remotely as well. > > If you want the text output that is shown, you can use the > printed-buffer-chunk command. If you want to know what slots a chunk has, > then you would use the chunk-filled-slots-list command, and to get the > value of a slot in the chunk you would use the chunk-slot-value command. > > Dan > On 7/26/20 1:24 PM, Arun Krishna wrote: > > Hi, > My java application uses RPC to communicate with ACT-R. > Mainly uses the evaluate method and is working in many cases. > But buffer-chunk command is not working via evaluate eventhough > buffer-read command works. > > For example > ----------------------------------------------- > JSONArray jsonArray3 = new JSONArray(); > jsonArray3.put("{\"method\":\"evaluate\","); > jsonArray3.put("\"params\":"); > jsonArray3.put("[\"buffer-read\",\"maneuver-left\",\"goal\"],"); > jsonArray3.put("\"id\":1}"); > > Works and I am getting proper out put > {"result": ["D-0"], "error": null, "id": 1} > -------------------------------------------------- > > > ------------------------------------------------------------ > But when I use buffer-chunk > JSONArray jsonArray3 = new JSONArray(); > jsonArray3.put("{\"method\":\"evaluate\","); > jsonArray3.put("\"params\":"); > jsonArray3.put("[\"buffer-chunk\",\"maneuver-left\",\"goal\"],"); > jsonArray3.put("\"id\":1}"); > > I am getting {"result": [["D-0"]], "error": null, "id": 1} which does not > include the contents in the > goal buffer. > > When I use the same command via ACT-R command line, buffer-chunk works and > gives complete output > (buffer-chunk goal) > GOAL: D-0 > D-0 > STATE3 "active" > > (D-0) > ------------------------------------------------------------------ > > Can you please let me know how to get the full contents in the > goal/retrieval buffer via RPC? > (Note: I am using ACT-R\examples\connections\Java\simpleACTRtest.java for > testing) > > Thanks & Regards, > Arun > > > > _______________________________________________ > ACT-R-users mailing listACT-R-users at act-r.psy.cmu.eduhttps://mailman.srv.cs.cmu.edu/mailman/listinfo/act-r-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: