INSTALLING THNET ================================================================== To obtain a copy of the THNET package (including ACME, ARCS, and ECHO), first download it: % ftp ftp.cognet.ucla.edu (or ftp 128.97.8.19) When connected, login as anonymous, and type in your login name for the password. Then: ftp> cd pub/THNET ....... ftp> get thnet.tar ....... ftp> quit Next, extract the THNET archive on your side: % tar -xf thnet.tar cd into THNET and check out the directory. It should contain the following files: REPS graphics.lisp misc.lisp semantics_s.lisp acme.lisp init.lisp patches.lisp staracme.bak acme.script links.lisp run.lisp staracme.lisp analogy.lisp loadacme.lisp semantics.lisp struc.lisp arcs.lisp loadarcs.lisp semantics_a.lisp thnet.man arcs.script loadecho.lisp semantics_e.lisp transfer.lisp arcsutil.lisp loadnet.lisp semantics_i.lisp util.lisp echo.lisp loadsem.lisp semantics_n.lisp REPS is a directory which contains some example representation files for ACME, ARCS, and ECHO. thnet.man is the THNET manual- it describes THNET's commands. Some other files worth mentioning are init.lisp, which contains global variables for the simulations (such as default excitation, number of cycles, etc.), graphics.lisp, which contains a set of graphics routines (available only on Suns- see thnet.man), and semantics.lisp, which loads a semantic network for ARCS. Another interesting (but probably useless) file is staracme.lisp, an implementation of ACME for the Connection Machine. To install THNET, you must edit the global variable *code-pathname* at the beginnig of loadacme.lisp, loadarcs.lisp, loadecho.lisp, and loadnet.lisp. These are the respective "loader files" for ACME, ARCS, and ECHO. Next, enter lisp, and compile THNET: % lisp .... > (load "loadnet") .... > (compile_thnet) .... > (quit) Now you are ready to use THNET. To use any of the programs, just enter lisp, load the appropriate loader file, and execute the simulation by calling the appropriate lisp routines. The file "example1.lisp" in the REPS directory is a good file to get started on: % lisp ..... > (load "loadacme") ..... > (load "REPS/example1") ..... > (make_struc1) Structure made: CAPTURE-FORTRESS T > (make_struc2) Structure made: DESTROY-TUMOR T > (constraint_map 'capture-fortress 'destroy-tumor) Constructing analogical mapping between CAPTURE-FORTRESS and DESTROY-TUMOR Using special unit. Constructing constraint network for CAPTURE-FORTRESS and DESTROY-TUMOR Making units and excitatory links for field GOALS ... Units made: 7 Symmetric links made: 11 Making units and excitatory links for field START ... Units made: 34 Symmetric links made: 64 Total number of units made: 41 Total number of symmetric excitatory links: 75 Making inhibitory links ... Symmetric inhibitory links made: 69 Total symmetric links made: 144 > (run_hyp_net 'acme) ............. Calculating the best mappings after 61 cycles. Best mapping of SAFE is ALIVE. 0.014361027371173467 Best mapping of CAPTURE is DESTROY. 0.69418597365052 Best mapping of GO_DOWN is OUTSIDE. 0.6145793567851688 Best mapping of LEAD_TO is SURROUND. 0.670924080897049 Best mapping of FORTRESS is TUMOR. 0.5795209455742865 Best mapping of ROADS is TISSUE. 0.5201890672020303 Best mapping of ARMY is RAY-SOURCE. 0.5768306035377417 Best mapping of CF7 is DT7. 0.014361027371173467 Best mapping of CF6 is DT6. 0.69418597365052 Best mapping of CF5 is DT5. 0.6145793567851688 Best mapping of CF4 is DT4. 0.670924080897049 Best mapping of CF3 is DT3. 0.5795209455742865 Best mapping of CF2 is DT2. 0.5201890672020303 Best mapping of CF1 is DT1. 0.5768306035377417 Best mapping of OBJ_FORTRESS is OBJ_TUMOR. 0.8331106658054727 Best mapping of OBJ_ROADS is OBJ_TISSUE. 0.6038388556861514 Best mapping of OBJ_ARMY is OBJ_RAY. 0.8194258571707753 NIL > (quit) Happy THNETing........ If you have questions contact holyoak@cognet.ucla.edu or melz@isi.edu