File size: 91,194 Bytes
6fa4bc9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 |
{
"paper_id": "I11-1027",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T07:31:27.653903Z"
},
"title": "Natural Language Programming Using Class Sequential Rules",
"authors": [
{
"first": "Sujay",
"middle": [],
"last": "Cohan",
"suffix": "",
"affiliation": {
"laboratory": "Aiaioo Labs Bangalore",
"institution": "",
"location": {
"country": "India"
}
},
"email": "cohan@aiaioo.com"
},
{
"first": "",
"middle": [],
"last": "Carlos",
"suffix": "",
"affiliation": {
"laboratory": "Aiaioo Labs Bangalore",
"institution": "",
"location": {
"country": "India"
}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "This paper presents a system for Natural Language Programming using Class Sequential Rules (CSR). The system recognizes a number of procedural primitives and operators. The domain of the system is presently limited to the world of numbers and operations on numbers. We evaluate the effectiveness of CSRs at the task of Natural Language Programming using an annotated corpus of programming instructions in natural language, achieving a precision and recall of 85% and 64% respectively. We also compare the performance of a system trained on annotated data with that of a system using handcrafted rules.",
"pdf_parse": {
"paper_id": "I11-1027",
"_pdf_hash": "",
"abstract": [
{
"text": "This paper presents a system for Natural Language Programming using Class Sequential Rules (CSR). The system recognizes a number of procedural primitives and operators. The domain of the system is presently limited to the world of numbers and operations on numbers. We evaluate the effectiveness of CSRs at the task of Natural Language Programming using an annotated corpus of programming instructions in natural language, achieving a precision and recall of 85% and 64% respectively. We also compare the performance of a system trained on annotated data with that of a system using handcrafted rules.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Since the early days of computing, there have been those who have longed for a programming system wherein the use of a formal symbolism for programming was not strictly necessary. Though high level languages greatly improve the ease of programming, it might be easier for a user to be allowed to communicate with an application in a natural language rather than a formal language with a specialized syntax. Moreover, a significant fraction of the population of the world is not conversant in the English language or remains unfamiliar with the Roman alphabet, and for them, a system for programming in their native tongues would be of great benefit. It is also easy to foresee a future where speech recognition systems are so accurate and robust that users might seek to provide instructions by means of speech to their computers. Thus there seems to be a need for algorithms capable of accepting commands and programming instructions in unconstrained human languages.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Arity Example if 2 or 3 If x is 2, say \"Hi\" unless 2 Exit unless x is 2 while 2 While x \u2264 2 . . . until 2 Till x is 2 add 1 to x continuation 1 Also, increment y assignment 2 Let x be 1 imperatives 0 to \u221e Display x questions 1 What is y? y/n questions 1 Is x equal to 2? We present a Natural Language Programming system capable of accepting programming commands in a natural language, executing them and returning any requested results to the user. It is limited to the domain of real numbers and can be used to write programs to compute the values of various functions of real numbers, or to generate different number series. The system can recognize nine broad categories of procedural primitives, some of which are conditionals, loops, assignments and function calls (imperatives). The complete list of types of procedural primitives with their arity and examples of usage can be found in Table 1 .",
"cite_spans": [],
"ref_spans": [
{
"start": 892,
"end": 899,
"text": "Table 1",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "We also present and evaluate a novel approach for processing user instructions. Instead of the traditional programming language approach of using a parser to produce a parse tree starting from modules or blocks of programming instructions, we first break up the programming instructions into sentences. We then use a short text classifier to first classify each resulting sentence into one of the categories of primitives listed above. Then we perform entity extraction to obtain as many contiguous and non-overlapping word subsequences as the arity of the primitive recognized, and then repeat the process with each of the word subse-quences. Our system is capable of learning to recognize programming instructions belonging to the categories described above from an annotated corpus. It can also use manually crafted rules for the recognition of commands from sentences in a natural language.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "The output of the above process is a semantic parse of the program. The semantic parses of consecutive sentences are joined into blocks if continuations are indicated. For example, if the first line is \"If x is 3, display x.\" and the second line is \"Also, increment x.\", the first line is semantically parsed into \"x = 3 =>print(x).\" and the second line is parsed to \"also(++x).\" Then the two lines are combined to get \"x = 3 =>print(x) && ++x.\" The semantic parse is also an interlingua representation that can be stored, used to perform an automatic translation of the program from one language to another, or executed to obtain the output. Mihalcea et al (2006) distinguish the two complementary programming tasks of description and proceduralization. The present paper deals with proceduralization, the process of constructing procedures out of steps, blocks, conditionals and loops. Of the procedural primitives listed in Table 1, the first two are conditional statement primitives; the next two are loop primitives; the fifth is used to construct blocks of statements by agglutination; the remainder are primitives for steps.",
"cite_spans": [
{
"start": 643,
"end": 664,
"text": "Mihalcea et al (2006)",
"ref_id": "BIBREF14"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "Imperatives tend to be function calls or commands that result in an action or change. Examples of imperatives include calls to \"display the value of x\" and \"Go to the step marked 'Subroutine 1'.\"",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "Wh-questions and yes/no questions have the effect of displaying the value of a variable, literal or expression. Yes/no questions are distinguished from wh-questions because their argument is constrained to be a boolean expression whereas whquestions can refer to non-boolean variables, literals and expressions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "The set of expressions supported by the system is listed in Table 2 . It will be observed that many commonly used mathematical operations like exponentiation and logarithms are not included in the list. The list does however include the arithmetic and relational operators that have their own keywords in the C, C++ and Java programming languages. At present, only two of the logical operators, namely or and and, are supported.",
"cite_spans": [],
"ref_spans": [
{
"start": 60,
"end": 67,
"text": "Table 2",
"ref_id": "TABREF2"
}
],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "The rest of this paper is organized as follows: Section 2 presents related work on the topic of natural language programming. The annotated corpus used to evaluate the system and the annotation guidelines for the same are presented in Section 3. Section 4 describes the entity recognition algorithms in some detail. The entity recognition systems are evaluated in Section 5 on the annotated corpus, and the novel approach of training a natural language programming system from annotated text is compared with the approach of using manually crafted rules. The conclusions and future directions are presented in Section 6.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Type",
"sec_num": null
},
{
"text": "Attempts to develop natural language programming systems have been made in the past, and a working prototype called \"NLC\" was described by Ballard and Biermann (1979) . NLC was capable of accepting English commands in the imperative mood. It did not accept declaratives and interrogatives. Each input was required to begin with an imperative verb. However, NLC was capable of dealing with pronominal references, like the ones in the command \"Put the average of the first four entries in that row into its last entry\" and with procedure definitions and loops, though not with conditionals. A sample NLC program from Ballard and Biermann (1979) is provided in Table 3 . An example of a looping instruction in NLC is the last line of the sample program that instructs the computer to repeat the preceding steps over other rows of the matrix under consideration. Biermann et \"Choose a row in the matrix\" \"Divide its last entry by 3\" \"Repeat for the other rows\" Take the array [3, 5, 7, 4, 6, 2, 1] . Count from one up to the size of the array: Go over the array from the beginning to the end minus the counter: If the current element is bigger than the following element then exchange the current element with the following element. Print \"After: \" and print the array. al (1983) also described attempts to provide commands to the NLC system using the Nippon Electric DP-200 Connected Speech Recognizer. Knoell and Mezini (2006) described a programming language called Pegasus that lets users program using plain German or English statements. Pegasus uses a handcrafted phrase structure grammar to parse English or German language commands and convert them into an intermediate representation. The intermediate representation is turned into a Java program using a dictionary of code snippets or translated into a different language. Knoell and Mezini (2006) did not perform an evaluation of the system on a corpus of natural language programming commands. A sample program written in the Pegasus language (see Table 4 ) was provided in the paper to show what a Pegasus program to sort an array of numbers would look like.",
"cite_spans": [
{
"start": 139,
"end": 166,
"text": "Ballard and Biermann (1979)",
"ref_id": "BIBREF2"
},
{
"start": 627,
"end": 642,
"text": "Biermann (1979)",
"ref_id": "BIBREF2"
},
{
"start": 972,
"end": 975,
"text": "[3,",
"ref_id": null
},
{
"start": 976,
"end": 978,
"text": "5,",
"ref_id": null
},
{
"start": 979,
"end": 981,
"text": "7,",
"ref_id": null
},
{
"start": 982,
"end": 984,
"text": "4,",
"ref_id": null
},
{
"start": 985,
"end": 987,
"text": "6,",
"ref_id": null
},
{
"start": 988,
"end": 990,
"text": "2,",
"ref_id": null
},
{
"start": 991,
"end": 993,
"text": "1]",
"ref_id": null
},
{
"start": 1400,
"end": 1424,
"text": "Knoell and Mezini (2006)",
"ref_id": "BIBREF8"
},
{
"start": 1829,
"end": 1853,
"text": "Knoell and Mezini (2006)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [
{
"start": 658,
"end": 665,
"text": "Table 3",
"ref_id": "TABREF3"
},
{
"start": 2006,
"end": 2013,
"text": "Table 4",
"ref_id": "TABREF4"
}
],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Objections to Natural Language Programming have been recorded, most notably by Dijkstra (1978) in an article titled 'On the Foolishness of \"Natural Language Programming\"'. Knoell and Mezini (2006) on the other hand opine that a system capable of dealing with both, formal symbolism and natural text, might be better than one that only understands either.",
"cite_spans": [
{
"start": 79,
"end": 94,
"text": "Dijkstra (1978)",
"ref_id": "BIBREF5"
},
{
"start": 172,
"end": 196,
"text": "Knoell and Mezini (2006)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "A system for generating program skeletons from the text of programming assignments like the one shown in Table 5 was described by Mihalcea et al (2006) . The system did not attempt to generate an executable program. Pane and Myers (2000) studied how nonprogrammers would describe solutions to prob-Write a program to generate 1000 numbers between 0 and 99 inclusive. You should count how many times each number is generated and write these counts out to the screen. lems arising within a Pacman game, and from their study, Pane and Myers (2001) proposed several principles of usability for developing a programming system for children. examine mixed-initiative dialog as a way of more precisely ascertaining user intention with respect to programming commands, again in the context of the Pacman game.",
"cite_spans": [
{
"start": 130,
"end": 151,
"text": "Mihalcea et al (2006)",
"ref_id": "BIBREF14"
},
{
"start": 216,
"end": 237,
"text": "Pane and Myers (2000)",
"ref_id": "BIBREF15"
},
{
"start": 523,
"end": 544,
"text": "Pane and Myers (2001)",
"ref_id": "BIBREF16"
}
],
"ref_spans": [
{
"start": 105,
"end": 112,
"text": "Table 5",
"ref_id": "TABREF5"
}
],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The present work is also related to the area of natural language understanding. Shapiro 2001describes a system that understands user statements about their beliefs about the world. This is in essense a form of descriptive programming, similar to that described in Lieberman and Liu (2005) who explore the possibility of using natural language descriptions as a representation for programs, and describe a system called \"Metafor\" capable of generating Python scaffolding (referred to as the visualization code) from them, though not fully specified programs.",
"cite_spans": [
{
"start": 264,
"end": 288,
"text": "Lieberman and Liu (2005)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "There has been, to our knowledge, no prior attempt to measure the performance of a natural language programming system using a corpus of annotated programming statements in a natural language. There has also not been, to our knowledge, any prior attempt to use entity recognition on natural language commands to generate fully specified programs. We also believe that this is the first attempt to learn patterns of language for programming from an annotated corpus.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Order Survey Question 1 How would you say \"x = 2\" in English? 2 How would you say \"x != 2\" in English? 4",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "How would you say \"x \u2264 2\" ? 6",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "How would you say \"x \u2265 2\" ? 9",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "How would you say \"x multiplied by 2\" ? 10 How would you say \"x / 2\" ? Table 6 : Survey Questions.",
"cite_spans": [],
"ref_spans": [
{
"start": 71,
"end": 78,
"text": "Table 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The sentences were collected by setting up an online survey on a website, sending out requests over social networks and using email. All the questions were presented to the user at the same time, and the instructions at the top said:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Please type into the box under each of the following, one way of saying the same thing in English and click on the Submit link below it. You can repeatedly enter different phrases if you can think of many ways of saying the same thing. For example, to communicate the idea of x = y, you might say x and y are equal or assign the value y to x.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The survey consisted of a total of thirty questions, and took approximately thirty minutes to complete. All users were presented with the same set of questions and in the same order. The answers provided by previous users were not made available to subsequent survey takers.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "A total of 3,517 sentences 2 was collected over a period of one month.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The annotation procedure involved recategorizing the submissions according to the procedural primitive or expression they most closely matched and marking the entity spans according to the following annotation rules:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "\u2022 Conditional primitives: A complex sentence specifying a condition for performing an action is a conditional statement. It is classified as an if conditional statement if it is conditional upon a positive outcome of the conditioning expression and the action to be taken does not undeniably suggest repeatedly checking the conditioning expression as a precondition for performing the action. It is classified as an unless conditional statement if it is conditional upon a negative outcome of the conditioning expression.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "\u2022 Loop primitives: A complex sentence specifying a condition that is repeatedly evaluated for performing an action until it is satisfied or denied is a loop statement. It is classified as a while loop statement if it is conditional upon a positive outcome of the conditioning expression. It is classified as an until loop statement if it is conditional upon a negative outcome of the conditioning expression.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "\u2022 Operations with side effects: Requests to add a value to a variable or to subtract a value from a variable are considered increment and decrement operations if the value of the variable would normally be considered to have changed at the end of the operation. For example, \"Add 2 to x\" would be considered an increment operation, whereas \"Add x to 2\" would not. The latter would be considered an addition operation and not an increment operation since it does not suggest a change in the value of x whereas the former does.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "\u2022 Operations without side effects: Sentences in the indicative mood, informative clauses and phrases are classified as operations that do not have side effects, if they are not expected to alter the value of a variable. For example, \"x added to 5\" and \"x by 5\" have no side effects.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "\u2022 Relational Operations: The classification of relational operations like less than depends on the order in which the parameters are supplied. So, \"x is less than 2\" and \"x is not greater than or equal to 2\" are both valid ways of representing \"x < 2\", but not \"2 is greater than x\". The last example is classified as a greater than operator.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "From the counts for the categories if conditional statement (95 before annotation and 118 after cleaup) and the unless conditional statement (64 before and 15 after) it appears to be the case that people strongly prefer using an if conditional statement to an unless conditional statement. Thus the number of test sentences in each of the categories is not balanced.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "As the system is intended for use as a multilingual teaching tool for students of computer programming, especially for those who do not possess a knowledge of English or the Roman alphabet, it was thought desirable to use a method of processing natural language programming instructions that would permit the rules for processing instructions to be learnt from annotated text in a number of possibly very different human languages.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "System Description",
"sec_num": "4"
},
{
"text": "One approach that seemed very promising was Class Sequential Rules (CSR) as described by Hu and Liu (2006) . CSRs have been shown to outperform existing methods at the task of associating opinions with product features . Moreover, CSRs can also be easily created by hand since they are a subset of cascading grammar rules such as those described by the Common Pattern Specification Language (CPSL) (Appelt, 1996) , which incidentally was developed as a language for specifying finite-state grammars for the purpose of information extraction. It was intuitively felt that a less powerful formalism might not only suffice for the task of processing programming commands, but also prove learnable from an annotated corpus.",
"cite_spans": [
{
"start": 89,
"end": 106,
"text": "Hu and Liu (2006)",
"ref_id": "BIBREF7"
},
{
"start": 398,
"end": 412,
"text": "(Appelt, 1996)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "System Description",
"sec_num": "4"
},
{
"text": "A Class Sequential Rule consists of a sequence of ordered tokens, that we indicate by the symbols i 1 . . . i n , for example, I = < i 1 i 2 i 3 >. A CSR matches a sentence only when each token in the CSR's token sequence matches a word token in the text under evaluation, in the right order. The CSR I will match a sentence s if and only if, in the sentence, there is a token s 3 that matches i 3 , and this token follows a token s 2 that matches i 2 . This second token must in turn follow a token s 1 that matches i 1 . For example, I will match the se-",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Class Sequential Rules",
"sec_num": "4.1"
},
{
"text": "quence < i 1 x 3 i 2 x 4 i 3 x 5 > but not < i 2 i 3 > or < i 3 i 2 i 1 >.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Class Sequential Rules",
"sec_num": "4.1"
},
{
"text": "CSRs like I can be used for classification as follows: a number of mutually exclusive CSRs are assigned to each class and used in conjunction with a priority or ordering scheme to resolve conflicts when CSRs from different classes match the input.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Class Sequential Rules",
"sec_num": "4.1"
},
{
"text": "CSRs can also have class labels c 1 . . . c n in their sequences. A class label can match zero or more tokens in the sentence. The tokens that class labels match represent entites in the sentences when Sequences < c 2 keab > < dc 2 kb > Table 7 : Sequence database. CSRs are used for entity extraction. For example, the CSR J =< i 1 c 4 i 2 c 5 i 3 > will match the sen-",
"cite_spans": [],
"ref_spans": [
{
"start": 237,
"end": 244,
"text": "Table 7",
"ref_id": null
}
],
"eq_spans": [],
"section": "Class Sequential Rules",
"sec_num": "4.1"
},
{
"text": "1 < c 2 > 2 < c 2 k > 3 < c 2 kb >",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "tence < i 1 x 3 x 4 i 2 i 3 >.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "The two class labels in J, namely c 4 and c 5 will at the same time match the sub-sequence < x 3 x 4 > and the empty sequence <> respectively. As you can see, the actual matching is performed by the sequence tokens i 1 . . . i n . The class labels pick up the tokens in between.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "If two class labels follow one another in quick succession (are not separated by a token in the sequence), for example < i 1 c 2 c 3 i 4 >, the extents of their spans are not well defined. The tokens that class labels match can be thought of as entities. Thus, in addition to classification, CSRs can be said to be capable of performing entity extraction.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "The task of understanding a programming command given in natural language can be broken down into the two sub-tasks that CSRs perform: a) classifying the command into one of several categories of commands; and b) extracting the arguments for further processing. Both tasks can be performed by CSRs in a single step.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "The algorithm for mining CSRs used in the present work is described in more detail in Hu and Liu (2006) . Using the algorithm, it is possible to mine sequences with a given minimum support. For instance, given two sequences such as those shown in Table 7 , and a minimum required support of 2, it is possible to extract all the patterns of length 1, 2 and 3 indicated in Table 8 . CSRs are only a special case of sequential patterns. With CSRs, the sequential patterns mined are text tokens. Some algorithms for sequential pattern mining have been studied in Agrawal and Srikant (1995) .",
"cite_spans": [
{
"start": 86,
"end": 103,
"text": "Hu and Liu (2006)",
"ref_id": "BIBREF7"
},
{
"start": 559,
"end": 585,
"text": "Agrawal and Srikant (1995)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [
{
"start": 247,
"end": 254,
"text": "Table 7",
"ref_id": null
},
{
"start": 371,
"end": 378,
"text": "Table 8",
"ref_id": "TABREF6"
}
],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "Other pattern exraction concepts closely related to CSRs include the surface patterns described by Table 9 . Ravichandran and Hovy (2002) , Hearst (1992) , Snow et al (2005) and Lin and Pantel (2001) .",
"cite_spans": [
{
"start": 109,
"end": 137,
"text": "Ravichandran and Hovy (2002)",
"ref_id": "BIBREF17"
},
{
"start": 140,
"end": 153,
"text": "Hearst (1992)",
"ref_id": "BIBREF6"
},
{
"start": 156,
"end": 173,
"text": "Snow et al (2005)",
"ref_id": "BIBREF19"
},
{
"start": 178,
"end": 199,
"text": "Lin and Pantel (2001)",
"ref_id": "BIBREF12"
}
],
"ref_spans": [
{
"start": 99,
"end": 106,
"text": "Table 9",
"ref_id": "TABREF7"
}
],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "Sequences Also { x = 2 }. Also, { x = 2 }. Also { if x = 3 , ++x }.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "In the course of developing a hand-crafted set of rules for natural language programming using CSRs, it was observed that the discriminative power of CSRs did not always suffice. For example, it was observed that CSRs could not accurately identify the entity spans in the three sentences listed in Table 9 .",
"cite_spans": [],
"ref_spans": [
{
"start": 298,
"end": 305,
"text": "Table 9",
"ref_id": "TABREF7"
}
],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "The reason is that the two rules in Table 10 are needed to match the spans in the first two sentences and these are ordered to fire one behind the other as shown in the table. Now, however, the first rule \"Also , EXPRESSION .\" incorrectly picks out the single entity span in the third sentence owing to the comma in the middle of the sentence, and there is no way to rectify the problem using a different number of CSRs or changing the ordering.",
"cite_spans": [],
"ref_spans": [
{
"start": 36,
"end": 44,
"text": "Table 10",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "Thus, a family of rules with more discriminative power was needed. We attempted to extend the concept of CSRs to give them more discriminative power, as described in the next subsection.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Sequences",
"sec_num": null
},
{
"text": "The extension that solved the problem described in the preceding section was that of allowing each token in the CSR to be an n-gram. The rules presented in Table 11 contain a special term \"NONE\" which indicates an n-gram constraint as follows:",
"cite_spans": [],
"ref_spans": [
{
"start": 156,
"end": 164,
"text": "Table 11",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Extended Class Sequential Rules",
"sec_num": "4.2"
},
{
"text": "Order Extended CSRs 1 Also NONE , EXPRESSION . 2 Also EXPRESSION . \u2022 When the term \"NONE\" appears between two tokens, the tokens they match in a sentence must be consecutive tokens. This is equivalent to making tokens on either side of \"NONE\" part of an n-gram. In the example in Table 11 , the first rule can only match sentences where a comma immediately follows the word \"Also.\"",
"cite_spans": [],
"ref_spans": [
{
"start": 280,
"end": 288,
"text": "Table 11",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Extended Class Sequential Rules",
"sec_num": "4.2"
},
{
"text": "\u2022 The term \"NONE\" can also appear at the beginning of an extended CSR to indicate that the following token must appear at the beginning of the matched sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Extended Class Sequential Rules",
"sec_num": "4.2"
},
{
"text": "\u2022 Similarly, it can appear at the end of an extended CSR to indicate that the preceding token can only appear as the last token of a matched sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Extended Class Sequential Rules",
"sec_num": "4.2"
},
{
"text": "The set of rules in Table 11 will be seen to be able to match all three sentences' spans correctly.",
"cite_spans": [],
"ref_spans": [
{
"start": 20,
"end": 28,
"text": "Table 11",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Extended Class Sequential Rules",
"sec_num": "4.2"
},
{
"text": "The two parameters used to evaluate the suitability of a rule are support (the percentage the sentences belonging to a category that it correctly identifies as belonging to that category) and confidence (the percentage of matches of the rule that were right), which are analogous to precision and recall. Only those rules were selected whose support and confidence values on the training data both exceeded minimum thresholds. The selected rules were ordered as follows:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule Selection and Ordering",
"sec_num": "4.3"
},
{
"text": "\u2022 Rules whose accuracy of span matching exceeded the threshold were placed first.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule Selection and Ordering",
"sec_num": "4.3"
},
{
"text": "\u2022 Rules whose accuracy of span matching fell below the threshold followed.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule Selection and Ordering",
"sec_num": "4.3"
},
{
"text": "\u2022 Within the two categories described above, longer rules went ahead of shorter rules.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule Selection and Ordering",
"sec_num": "4.3"
},
{
"text": "The intermediate representation is a programming language that mimics the ambiguities of the language used in mathematics. The intermediate representation differs from a conventional programming language like Python in the following ways:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Intermediate Representation",
"sec_num": "4.4"
},
{
"text": "\u2022 Terms which tend to be ambiguous in natural language remain so in the intermediate representation. For example, the 'assignment operator' doubles as the 'equal to operator'",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Intermediate Representation",
"sec_num": "4.4"
},
{
"text": "Let x be 3. y is 9. What is x times y? While x is less than y, print x and then increment x. A program written in the present system would look like that in Table 12 .",
"cite_spans": [],
"ref_spans": [
{
"start": 157,
"end": 165,
"text": "Table 12",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Intermediate Representation",
"sec_num": "4.4"
},
{
"text": "The natural language programming system was evaluated against the annotated corpus described in Section 3. The manual rules used in the evaluation were developed and fixed prior to the start of corpus collection to avoid the introduction of biases through any knowledge of the corpus to be tested on.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation and Results",
"sec_num": "5"
},
{
"text": "The categories that were used in testing were equality, inequality, less than, greater than, less than or equal to, greater than or equal to, addition, subtraction, multiplication, division, increment, decrement, if, while, unless, until, print, conjunctive, disjunctive, divisible and continuation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Categories",
"sec_num": "5.1"
},
{
"text": "The categories in the corpus that were left out of the evaluation were as follows:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Categories",
"sec_num": "5.1"
},
{
"text": "\u2022 Three categories were left out because of the lack of manually crafted rules for those categories.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Categories",
"sec_num": "5.1"
},
{
"text": "\u2022 Of the three categories recognized as the print command by the manual rules, only one was retained for the experiment. \u2022 The modulus operator was left out of the evaluation because the manual rules treated the operator as 'modulus' whereas the question used in the survey used to develop the corpus had suggested that the operator was the 'absolute value' operator.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Categories",
"sec_num": "5.1"
},
{
"text": "The 3,000 sentences in the annotated corpus belong to 29 distinct categories of which 21 are used for evaluating the system. Support and confidence values of 0.0001 and 0.70 3 respectively were used during training (for rule discovery).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "Since we performed 3-fold cross validation, three sets of experiments were conducted for each of the following settings, for a total of 9 experiments in all:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "\u2022 Conventional CSRs (CSR-BL)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "\u2022 Extended CSRs (CSR-EX)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "\u2022 Manually crafted rules (CSR-Man) Setting Prec.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "Recall F1 CSR-Man 89.2 \u00b1 3.7 64.8 \u00b1 6.2 73.0 \u00b1 4 CSR-BL 85.7 \u00b1 4.5 65.3 \u00b1 5.9 73.1 \u00b1 4 CSR-EX 88.4 \u00b1 3.4 66.5 \u00b1 5.6 74.8 \u00b1 3 In all experiments, the Precision, Recall and F1 Score (the harmonic mean of Precision and Recall) were measured for each of the categories, as well as the overall accuracy of categorization. The Precision, Recall and F1 scores for the CSR-EX algorithm are presented in Table 13 . In the second column of the table is listed the number of sentences used in the test. This value in some cases drops to as low as fifteen sentences. The confidence intervals are rather high, making it difficult to draw comparisons between algorithms based on this data. The average of these scores for all categories is reported in Table 14 .",
"cite_spans": [],
"ref_spans": [
{
"start": 395,
"end": 403,
"text": "Table 13",
"ref_id": "TABREF0"
},
{
"start": 738,
"end": 746,
"text": "Table 14",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "The accuracy of entity span boundary detection is measured as follows: A recall-based score for correct span detection is computed by dividing the number of sentences with perfectly identified spans by the number of sentences in the category. This score is reported as the PSCS (percentage of sentences with correct spans). This score is similar to but not quite the same as the PCS (percentage of correct scopes) metric used in Councill et al (2010).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "The PSCS scores for the three algorithms are reported in Table 15 . We observe from the results that for the corpus the evaluation was performed on, there is no significant difference between the algorithms evaluated.",
"cite_spans": [],
"ref_spans": [
{
"start": 57,
"end": 65,
"text": "Table 15",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "The overall accuracy scores presented in Table 16 again reveal no significant differences between CSR-Man, CSR-BL and CSR-EX in their behaviour with respect to the data-set.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "Metric CSR-Man CSR-BL CSR-EX Acc.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "64.3 \u00b1 7 64.4 \u00b1 6 66.0 \u00b1 4 ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5.2"
},
{
"text": "This paper presents a system for Natural Language Programming capable of recognizing a number of categories of procedural programming instructions in a natural language. The system uses Class Sequential Rules to convert a natural language representation of a program into an intermediate representation that can be executed. The system is capable of using manually crafted rules or rules learnt from an annotated corpus. Since no corpus was available for evaluation of a system for Natural Language Programming, a corpus consisting of 3,000 sentences in twentynine categories (of which only twenty-one were used), was collected over the internet, cleaned, recategorized, annotated with entity spans and made publicly available.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions and Future Work",
"sec_num": "6"
},
{
"text": "Since the existing formalism of Class Sequential Rules (CSR-BL) was not powerful enough to tease certain sets of sentences apart into the right categories, an extension to Class Sequential Rules was proposed (CSR-EX) and implemented. Finally, the system was evaluated by three-fold cross validation using the corpus. Three settings of the system were tested: a) a setting where it used extended CSR-EX rules manually crafted before the collection of the corpus; b) a setting where it used CSR-BL rules learnt from the annotated corpus; and c) a setting where it used CSR-EX rules learnt from the annotated corpus. Precisions of around 85% and recalls of approximately 64% were measured with confidence intervals as large as 7%. The large confidence intervals make it impossible to establish if one of the approaches works better than the others with the present corpus and the present set of categories.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions and Future Work",
"sec_num": "6"
},
{
"text": "Future research could include an evaluation on a larger corpus, on more languages and on the system's ability to adapt to new domains. It would also be interesting to examine system accuracies with an increased number of categories covering more operations and functions. It might also be of interest to build a corpus of complete programs rather than individual sentences, to capture more variations in language.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions and Future Work",
"sec_num": "6"
},
{
"text": "Data SetAt the time of writing, there was no corpus available for evaluating the performance of a system for procedural programming in a natural language. Therefore, a corpus was developed 1 consisting of sentences in the English language, which the contributors of the sentences perceived to be commands that ought to invoke the programming primitives listed inTable 1or specify the expressions listed inTable 2.1 The annotated corpus can be downloaded from the URL http://www.aiaioo.com/corpora/vaklipi2011.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The number 3,517 included blank sentences and names as well. When these were removed, we were left with about 3,100 sentences of which we annotated 3,000.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "These values were manually chosen keeping in mind the small size of the corpus. The support threshold was chosen to be low enough to not affect rule selection. The confidence threshold was kept low enough to permit single failures (incorrect matches) from time to time.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "We are very grateful to all the volunteers who contributed to the corpus and to Govind Sharma, Kartik Asooja, Dr. Pushpak Bhattacharyya and the reviewers for helpful suggestions and references.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgments",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Mining Sequential Patterns",
"authors": [
{
"first": "Rakesh",
"middle": [],
"last": "Agrawal",
"suffix": ""
},
{
"first": "Ramakrishnan",
"middle": [],
"last": "Srikant",
"suffix": ""
}
],
"year": 1995,
"venue": "Proceedings of the Eleventh International Conference on Data Engineering IEEE Computer Society",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rakesh Agrawal and Ramakrishnan Srikant. 1995. Mining Sequential Patterns. In Proceedings of the Eleventh International Conference on Data Engi- neering IEEE Computer Society Washington, DC, USA. 1995.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "The Common Pattern Specification Language",
"authors": [
{
"first": "Douglas",
"middle": [
"E"
],
"last": "Appelt",
"suffix": ""
}
],
"year": 1996,
"venue": "Proceedings of a workshop on",
"volume": "",
"issue": "",
"pages": "23--30",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Douglas E. Appelt. 1996. The Common Pattern Spec- ification Language. In Proceedings of a workshop on held at Baltimore, Maryland. 1996, 23-30.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Programming in Natural Language",
"authors": [
{
"first": "W",
"middle": [],
"last": "Bruce",
"suffix": ""
},
{
"first": "Alan",
"middle": [
"W"
],
"last": "Ballard",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Biermann",
"suffix": ""
}
],
"year": 1979,
"venue": "NLC\" as a prototype. Proceedings of the 1979 annual conference",
"volume": "",
"issue": "",
"pages": "228--237",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bruce W. Ballard, Alan W. Biermann. 1979. Program- ming in Natural Language: \"NLC\" as a prototype. Proceedings of the 1979 annual conference. 1979, 228-237.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Interactive natural language problem solving: a pragmatic approach",
"authors": [
{
"first": "Alan",
"middle": [
"W"
],
"last": "Biermann",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Rodman",
"suffix": ""
},
{
"first": "Bruce",
"middle": [
"W"
],
"last": "Ballard",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Betancourt",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Bilbro",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Deas",
"suffix": ""
},
{
"first": "L",
"middle": [],
"last": "Fineman",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Fink",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Gilbert",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Gregory",
"suffix": ""
},
{
"first": "F",
"middle": [],
"last": "Heidlage",
"suffix": ""
}
],
"year": 1983,
"venue": "ANLC '83, Proceedings of the first conference on Applied Natural Language Processing",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alan W. Biermann, R. Rodman, Bruce W. Ballard, T. Betancourt, G. Bilbro, H. Deas, L. Fineman, P. Fink, K. Gilbert, D. Gregory, F. Heidlage. 1983. Inter- active natural language problem solving: a prag- matic approach. In ANLC '83, Proceedings of the first conference on Applied Natural Language Pro- cessing, 1983.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Whats Great and Whats Not: Learning to Classify the Scope of Negation for Improved Sentiment Analysis",
"authors": [
{
"first": "G",
"middle": [],
"last": "Isaac",
"suffix": ""
},
{
"first": "Ryan",
"middle": [],
"last": "Councill",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Mcdonald",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the Workshop on Negation and Speculation in Natural Language Processing",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Isaac G. Councill , Ryan McDonald , Leonid Ve- likovich. 2010. Whats Great and Whats Not: Learning to Classify the Scope of Negation for Im- proved Sentiment Analysis. In Proceedings of the Workshop on Negation and Speculation in Natural Language Processing (NeSp-NLP 2010). 2010.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "On the foolishness of \"Natural Language Programming",
"authors": [
{
"first": "W",
"middle": [],
"last": "Edsger",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Dijkstra",
"suffix": ""
}
],
"year": 1978,
"venue": "Program Construction",
"volume": "",
"issue": "",
"pages": "51--53",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Edsger W. Dijkstra. 1978. On the foolishness of \"Nat- ural Language Programming\". In Program Con- struction. 1978, 51-53.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Automatic Acquisition of Hyponyms from Large Text Corpora",
"authors": [
{
"first": "Marti",
"middle": [
"A"
],
"last": "Hearst",
"suffix": ""
}
],
"year": 1992,
"venue": "Proceedings of the 14th conference on Computational Linguistics",
"volume": "2",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Marti A. Hearst. 1992. Automatic Acquisition of Hy- ponyms from Large Text Corpora. In Proceedings of the 14th conference on Computational Linguistics - Volume 2. 1992.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Opinion Feature Extraction Using Class Sequential Rules",
"authors": [
{
"first": "Minqing",
"middle": [],
"last": "Hu",
"suffix": ""
},
{
"first": "Bing",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2006,
"venue": "AAAI Spring Symposium on Computational Approaches to Analyzing Weblogs",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Minqing Hu and Bing Liu. 2006. Opinion Feature Ex- traction Using Class Sequential Rules. AAAI Spring Symposium on Computational Approaches to Ana- lyzing Weblogs, Palo Alto, USA, March 2006.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Pegasus First Steps Toward a Naturalistic Programming Language",
"authors": [
{
"first": "Roman",
"middle": [],
"last": "Knoell",
"suffix": ""
},
{
"first": "Mira",
"middle": [],
"last": "Mezini",
"suffix": ""
}
],
"year": 2006,
"venue": "Companion to the 21st ACM SIG-PLAN symposium on Object-oriented programming systems, languages, and applications",
"volume": "",
"issue": "",
"pages": "542--559",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Roman Knoell and Mira Mezini. 2006. Pega- sus First Steps Toward a Naturalistic Programming Language. In Companion to the 21st ACM SIG- PLAN symposium on Object-oriented programming systems, languages, and applications. 2006, 542- 559.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Metafor: Visualizing stories as code",
"authors": [
{
"first": "Henry",
"middle": [],
"last": "Lieberman",
"suffix": ""
},
{
"first": "Hugo",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2005,
"venue": "10th International Conference on Intelligent User Interfaces",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Henry Lieberman and Hugo Liu. 2005. Metafor: Visu- alizing stories as code. In 10th International Con- ference on Intelligent User Interfaces. 2005.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Feasibility studies for programming in natural language",
"authors": [
{
"first": "Henry",
"middle": [],
"last": "Lieberman",
"suffix": ""
},
{
"first": "Hugo",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Henry Lieberman and Hugo Liu. 2006. Feasibil- ity studies for programming in natural language.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "End-User Development (Human-Computer Interaction Series",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lieberman, Paterno, Wulf (Eds.): End-User Devel- opment (Human-Computer Interaction Series Vol.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Discovery of inference rules for question-answering",
"authors": [
{
"first": "Dekang",
"middle": [],
"last": "Lin",
"suffix": ""
},
{
"first": "Patrick",
"middle": [],
"last": "Pantel",
"suffix": ""
}
],
"year": 2001,
"venue": "In Journal Natural Language Engineering archive",
"volume": "7",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dekang Lin and Patrick Pantel. 2001. Discovery of inference rules for question-answering. In Journal Natural Language Engineering archive Volume 7 Is- sue 4, Cambridge University Press New York, NY, USA. December 2001.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Opinion Observer: Analyzing and Comparing Opinions on the Web",
"authors": [
{
"first": "Bing",
"middle": [],
"last": "Liu",
"suffix": ""
},
{
"first": "Minqing",
"middle": [],
"last": "Hu",
"suffix": ""
},
{
"first": "Junsheng",
"middle": [],
"last": "Cheng",
"suffix": ""
}
],
"year": 2005,
"venue": "Proceedings of the 14th International World Wide Web conference (WWW-2005)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bing Liu, Minqing Hu and Junsheng Cheng. 2005. Opinion Observer: Analyzing and Comparing Opin- ions on the Web. In Proceedings of the 14th Inter- national World Wide Web conference (WWW-2005), Chiba, Japan. 2005.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "NLP (Natural Language Processing) for NLP (Natural Language Programming)",
"authors": [
{
"first": "Rada",
"middle": [],
"last": "Mihalcea",
"suffix": ""
},
{
"first": "Hugo",
"middle": [],
"last": "Liu",
"suffix": ""
},
{
"first": "Henry",
"middle": [],
"last": "Lieberman",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of CICLing",
"volume": "",
"issue": "",
"pages": "319--330",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rada Mihalcea and Hugo Liu and Henry Lieberman. 2006. NLP (Natural Language Processing) for NLP (Natural Language Programming). In Proceedings of CICLing. 2006, 319-330.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "The Influence of the Psychology of Programming on a Language Design: Project Status Report",
"authors": [
{
"first": "John",
"middle": [
"F"
],
"last": "Pane",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Brad",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Myers",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of the 12th Annual Meeting of the Psychology of Programmers Interest Group",
"volume": "",
"issue": "",
"pages": "193--205",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "John F. Pane and Brad A. Myers. 2000. The Influ- ence of the Psychology of Programming on a Lan- guage Design: Project Status Report. In Proceed- ings of the 12th Annual Meeting of the Psychology of Programmers Interest Group, A. F. Blackwell and E. Bilotta, Eds. Corigliano Calabro, Italy: Edizioni Memoria, April 10-13 2000, 193-205.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Studying the language and structure in non-programmers' solutions to programming problems",
"authors": [
{
"first": "John",
"middle": [
"F"
],
"last": "Pane",
"suffix": ""
},
{
"first": "Ann",
"middle": [],
"last": "Chotirat",
"suffix": ""
},
{
"first": "Brad",
"middle": [
"A"
],
"last": "Ratanamahatana",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Myers",
"suffix": ""
}
],
"year": 2001,
"venue": "In International Journal of Human-Computer Studies",
"volume": "54",
"issue": "2",
"pages": "237--264",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "John F. Pane, Chotirat Ann Ratanamahatana and Brad A. Myers. 2001. Studying the language and structure in non-programmers' solutions to programming problems. In International Journal of Human-Computer Studies Volume 54, Issue 2, February 2001, 237-264.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Learning Surface Text Patterns for a Question Answering System",
"authors": [
{
"first": "Deepak",
"middle": [],
"last": "Ravichandran",
"suffix": ""
},
{
"first": "Eduard",
"middle": [],
"last": "Hovy",
"suffix": ""
}
],
"year": 2002,
"venue": "Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Deepak Ravichandran and Eduard Hovy. 2002. Learn- ing Surface Text Patterns for a Question Answering System. In Proceedings of the 40th Annual Meet- ing of the Association for Computational Linguis- tics, 2002.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "The CASSIE Projects: An Approach to Natural Language Competence",
"authors": [
{
"first": "C",
"middle": [],
"last": "Stuart",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Shapiro",
"suffix": ""
}
],
"year": 1989,
"venue": "EPIA",
"volume": "",
"issue": "",
"pages": "362--380",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stuart C. Shapiro. 1989. The CASSIE Projects: An Approach to Natural Language Competence. EPIA, 1989, 362-380.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Learning Syntactic Patterns for Automatic Hypernym Discovery",
"authors": [
{
"first": "Rion",
"middle": [],
"last": "Snow",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Jurafsky",
"suffix": ""
},
{
"first": "Andrew",
"middle": [
"Y"
],
"last": "Ng",
"suffix": ""
}
],
"year": 2005,
"venue": "Advances in Neural Information Processing Systems 17",
"volume": "",
"issue": "",
"pages": "1297--1304",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rion Snow, Daniel Jurafsky, Andrew Y. Ng. 2005. Learning Syntactic Patterns for Automatic Hyper- nym Discovery. In Advances in Neural Information Processing Systems 17. 2005, 1297-1304.",
"links": null
}
},
"ref_entries": {
"TABREF0": {
"text": "Types of Procedural Primitives.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF2": {
"text": "Expressions.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF3": {
"text": "Sample NLC Program.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF4": {
"text": "Sample Pegasus Program.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF5": {
"text": "Programming Assignment.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF6": {
"text": "Extracted sequences with support 2.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF7": {
"text": "Three annotated sentences that demonstrate the inadequacy of CSRs for natural language programming.",
"content": "<table><tr><td colspan=\"2\">Order CSRs</td></tr><tr><td>1</td><td>Also , EXPRESSION .</td></tr><tr><td>2</td><td>Also EXPRESSION .</td></tr></table>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF8": {
"text": "Class Sequential Rules for the entity spans in",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF9": {
"text": "",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF10": {
"text": "Sample commands in our system.",
"content": "<table><tr><td>in many natural languages. This overloading</td></tr><tr><td>poses no problem since the right form can be</td></tr><tr><td>resolved from context.</td></tr><tr><td>\u2022 The intermediate representation also attempts to capture the logical operator priority of In-</td></tr><tr><td>dian languages. In many Indian languages,</td></tr><tr><td>it is not possible to set or-phrases as sub-</td></tr><tr><td>phrases of and-phrases. So, we have chosen</td></tr><tr><td>a priority order for operations in the interme-</td></tr><tr><td>diate representation that naturally maintains</td></tr><tr><td>the restriction.</td></tr></table>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF11": {
"text": "\u00b1 06 66.5 \u00b1 19 71.9 \u00b1 10 inequality 165 90.6 \u00b1 14 78.6 \u00b1 06 84.3 \u00b1 09 less than 151 66.8 \u00b1 10 88.4 \u00b1 07 76.8 \u00b1 08 \u2264 137 99.1 \u00b1 02 75 \u00b1 13 86.0 \u00b1 07 more than 158 76.6 \u00b1 08 83.1 \u00b1 06 79.6 \u00b1 02",
"content": "<table><tr><td>Category Cnt Precision</td><td>Recall</td><td>F1</td></tr><tr><td colspan=\"3\">equality 298 79.0 \u2265 132 92.9 \u00b1 05 80.8 \u00b1 13 86.5 \u00b1 09 addition 140 97.9 \u00b1 04 61.2 \u00b1 10 77.2 \u00b1 06 subtract 113 92.5 \u00b1 15 71.0 \u00b1 06 80.8 \u00b1 06 multiply 144 98.8 \u00b1 02 64.1 \u00b1 12 79.4 \u00b1 08 division 143 89.8 \u00b1 10 69.8 \u00b1 08 79.2 \u00b1 09 increment 136 92.5 \u00b1 08 57.3 \u00b1 08 72.8 \u00b1 08 decrement 131 96.9 \u00b1 06 23.5 \u00b1 15 46.7 \u00b1 15 if 118 84.2 \u00b1 05 96.0 \u00b1 08 89.8 \u00b1 04 while 61 92.1 \u00b1 02 88.0 \u00b1 12 89.8 \u00b1 11 unless 15 100 \u00b1 00 60.7 \u00b1 15 77.6 \u00b1 09 until 86 98.8 \u00b1 02 85.8 \u00b1 15 91.9 \u00b1 08 print 82 92.3 \u00b1 06 33.9 \u00b1 14 55.1 \u00b1 09 and 68 52.8 \u00b1 11 82.8 \u00b1 14 66.1 \u00b1 12 or 67 92.1 \u00b1 08 37.8 \u00b1 04 58.8 \u00b1 03 divisible 66 92.7 \u00b1 08 71.1 \u00b1 18 80.7 \u00b1 10 continue 48 78.3 \u00b1 23 22.1 \u00b1 11 40.0 \u00b1 05</td></tr><tr><td>\u2022 Two of the omitted categories were merely synomyms for boolean constants 'true' and</td><td/><td/></tr><tr><td>'false.'</td><td/><td/></tr></table>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF12": {
"text": "Evaluation of CSR-EX.",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF13": {
"text": "Categorization Evaluation.",
"content": "<table><tr><td colspan=\"2\">Metric CSR-Man</td><td>CSR-BL</td><td>CSR-EX</td></tr><tr><td>PSCS</td><td colspan=\"3\">52.4 \u00b1 9.1 50.2 \u00b1 8.4 49.7 \u00b1 8.6</td></tr><tr><td colspan=\"4\">Table 15: Entity Span Matching Evaluation.</td></tr></table>",
"html": null,
"type_str": "table",
"num": null
},
"TABREF14": {
"text": "",
"content": "<table/>",
"html": null,
"type_str": "table",
"num": null
}
}
}
} |