1

Topic: Programming natural languages, move?

Good afternoon One of these days viewed "Self-determination in a context of global calls" Pereslegina. He certainly strongly disappoints the last 5 years. But curious ideas in video are. As the programmer I marked a row of punctures, for example there are "computers" already are tested  and other yellow snow. But surprised another - Pereslegin (and with them possibly other academic brotherhood) till now wait "programming in a natural language" as some qualitative change in IT. If I can not fairly understand what tasks "programming in a natural language at all" can solve, since  the majority of programmers manage nasty to program and in the formal languages. Or it is banal  "the academic brotherhood" before syntax learning? And they to hope - here when it will be possible to talk to the computer in "human language" here then we  how many all good !

2

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: IQ> If I can not fairly understand what tasks "programming in a natural language at all" can solve, since  the majority of programmers manage nasty to program and in the formal languages. IQ> or it is banal  "the academic brotherhood" before syntax learning? And they to hope - here when it will be possible to talk to the computer in "human language" here then we  how many all good ! Matter is not in syntax. The basic distinction between  and a natural language consists that the text on  has exact value and is unambiguously interpreted. If at compilation there is an uncertainty for example, the same identifier can be resolved in two objects), and the compiler has no explicit rules of its resolution there will be an error. The text in a natural language has uncertainty which we resolve on the basis of a context and personal experience, applying for this purpose the intelligence. Result at this process , therefore often different people understand the same text differently. Operation of the programmer in the essence also consists in translating unsharp setting of the task on natural language in the unambiguous description of the decision on the formal.

3

Re: Programming natural languages, move?

Hello, 0x7be, you wrote: 0> Hello, IQuerist, you wrote: IQ>> If I can not fairly understand what tasks "programming in a natural language at all" can solve, since  the majority of programmers manage nasty to program and in the formal languages. IQ>> or it is banal  "the academic brotherhood" before syntax learning? And they to hope - here when it will be possible to talk to the computer in "human language" here then we  how many all good ! 0> Matter is not in syntax. The basic distinction between  and a natural language consists that the text on  has exact value and is unambiguously interpreted. If at compilation there is an uncertainty for example, the same identifier can be resolved in two objects), and the compiler has no explicit rules of its resolution there will be an error. The text in a natural language has uncertainty which we resolve on the basis of a context and personal experience, applying for this purpose the intelligence. Result at this process , therefore often different people understand the same text differently. 0> Operation of the programmer in the essence also consists in translating unsharp setting of the task on natural language in the unambiguous description of the decision on the formal. It is clear if earlier for "programming", moreover, often meant to solve any task numerical methods, today to "program"  already means creation of conventional system/appendix that  in itself already by and large is not the "program" task and it is probable with IOT this trend qualitatively to amplify... Where here a place for "programming in a natural language"? PS  we see absolutely other trend - intuitivism of interfaces which gives the chance to unprepared users to use difficult enough systems. But it  goes in the full section with "programming in a natural language".

4

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: IQ> It is clear if earlier for "programming", moreover, often meant to solve any task numerical methods, today to "program"  already means creation of conventional system/appendix that  in itself already by and large is not the "program" task and it is probable with IOT this trend qualitatively to amplify... Where here a place for "programming in a natural language"? And figs it knows. Machine programming under the requirements formulated in a natural language? IQ> PS  we see absolutely other trend - intuitivism of interfaces which gives the chance to unprepared users to use difficult enough systems. But it  goes in the full section with "programming in a natural language". I do not know. To take, for example, Siri. When the request in a natural language is sent to it is unless not its programming on specific action? On the other hand, the language understood Siri is only some restricted subset of a natural language, and a context in which it is interpreted too is extreme narrow. It turns out such DSL

5

Re: Programming natural languages, move?

Hello, 0x7be, you wrote: 0> Hello, IQuerist, you wrote: IQ>> It is clear if earlier for "programming", moreover, often meant to solve any task numerical methods, today to "program"  already means creation of conventional system/appendix that  in itself already by and large is not the "program" task and it is probable with IOT this trend qualitatively to amplify... Where here a place for "programming in a natural language"? 0> And figs it knows. Machine programming under the requirements formulated in a natural language? We that with you understand development process entirely, and what part in this process is occupied with activity "formalization of requirements"? IQ>> PS  we see absolutely other trend - intuitivism of interfaces which gives the chance to unprepared users to use difficult enough systems. But it  goes in the full section with "programming in a natural language". 0> I do not know. To take, for example, Siri. When the request in a natural language is sent to it is unless not its programming on specific action? 0> on the other hand, the language understood Siri is only some restricted subset of a natural language, and a context in which it is interpreted too is extreme narrow. 0> It turns out such DSL Imho is not present... In my opinion it simply part of the interface, after all the new behavior does not appear.

6

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: 0>> And figs it knows. Machine programming under the requirements formulated in a natural language? IQ> we that with you understand development process entirely, and what part in this process is occupied with activity "formalization of requirements"? , all. Formalization comes to an end with the code - the unique formal description of system. Requirements, let even very detailed, written down in a natural language, all the same have uncertainty. IQ> Imho is not present... In my opinion it simply part of the interface, after all the new behavior does not appear. , let's up still. Course SICP defines a programming language as set of three components: 1. Primitives 2. Combination means. 3. Abstraction means.  allows to execute some primitives. If suddenly there will be a possibility to combine some operations in one scenario and to abstract it behind some new phrase, and to do it too by a voice, it turns out ? In my judgement - yes. Whether there will be it a natural language - , is faster - is not present

7

Re: Programming natural languages, move?

Hello, 0x7be, you wrote: 0> Hello, IQuerist, you wrote: 0>>> And figs it knows. Machine programming under the requirements formulated in a natural language? IQ>> we that with you understand development process entirely, and what part in this process is occupied with activity "formalization of requirements"? 0> , all. Formalization comes to an end with the code - the unique formal description of system. Requirements, let even very detailed, written down in a natural language, all the same have uncertainty. In the modern projects with the code comes to an end very much and very big group  IQ>> is not present Imho... In my opinion it simply part of the interface, after all the new behavior does not appear. 0> Okej, let's up still. Course SICP defines a programming language as set of three components: 0> 1. Primitives 0> 2. Combination means. 0> 3. Abstraction means. 0> Siri allows to execute some primitives. If suddenly there will be a possibility to combine some operations in one scenario and to abstract it behind some new phrase, and to do it too by a voice, it turns out ? In my judgement - yes. Whether there will be it a natural language - , is faster - there would be no That the programmer did not invent, as a result it turns out

8

Re: Programming natural languages, move?

Hello, , you wrote: Hello, IQuerist, you wrote: IQ>> PS  we see absolutely other trend - intuitivism of interfaces which gives the chance to unprepared users to use difficult enough systems. But it  goes in the full section with "programming in a natural language". Housewives will soon program in a natural language. But whether it will be possible to consider it as programming? With ours  the points of view - are not present more likely, with home - likely - yes. A question  basic enough... They will be engaged in "development" or to use possibilities of the advanced interfaces?  you and  in programmers write down

9

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: the Entertaining fact. In Japanese language practically  application of a word "is not present". And generally, an imperative mood. Once it was, but only between the master and the slave. As a result it is impossible to tell in a bar, give to me . It is necessary to tell something of type " to have to me it is desirable". Frequently meetings of programmers are led in English even if Japanese participate only. And to program on Japanese - it is impossible.

10

Re: Programming natural languages, move?

I think, "programming in a natural language" is the most easier for implementing in the form of formalizing dialogue with the computer. That is, the person will try to formulate requirements to the program as can, and the computer will ask specifying questions, trying to eliminate all ambiguities until unambiguity of the next requirement will not be reached. Most likely, instead of the program code the formal description of the requirements will appear, able to be unambiguously displayed in the text description in a natural language (its subset if is more exact). The next stage of programming - depositing of the next requirement - will be considered completed, whether when the computer reads the formal description of this requirement and of its questions "All is true? Specifications are not present?" The person answers accordingly "Yes" and "is not present". In some measure and now, when the programmer tries to feed to the compiler the incorrect code (not up to the end formalized requirements), between the programmer and the compiler there is certain "dialogue" (consisting of error messages and attempts to correct the code), coming to an end with the full formalization. It is necessary to translate this dialogue to a natural language. As ghost effect, on a course of operation the programmer will master naturally singularities of the formal language of the description of requirements and will try to speak with the computer at once on it, without using inappropriate in a specific case of a figure of a natural language. So it can enter/dictate requirements with smaller number of errors, reducing a stage of formalizing dialogue.

11

Re: Programming natural languages, move?

Hello, L_G, you wrote: L_G> That is, the person will try to formulate requirements to the program as can, and the computer will ask specifying questions, trying to eliminate all ambiguities until unambiguity of the next requirement will not be reached. And whence it is known, what it generally can be reached? Here people cannot force to formulate normally customers of the requirement, and you about the machine.

12

Re: Programming natural languages, move?

Hello, Ops, you wrote: Ops> And whence it is known, what it generally can be reached? Here people cannot force to formulate normally customers of the requirement, and you about the machine. Can. It is necessary to add only "default". For example, "write me PrivetMir" - deduces "Greetings of Patterns" white letters on a black background in a screen corner. If it is not pleasant to the user, he starts to add specifications.

13

Re: Programming natural languages, move?

Hello, Real 3L0, you wrote: R3> Can. It is necessary to add only "default". R3> For example, "write me PrivetMir" - deduces "Greetings of Patterns" white letters on a black background in a screen corner. If it is not pleasant to the user, he starts to add specifications. It is all well for primitive tasks. And how about "create to me the program of document circulation for my firm"?

14

Re: Programming natural languages, move?

Ops> It is all well for primitive tasks. And how about "create to me the program of document circulation for my firm"? What for? "Register the document", "Find the document", "Transfer to Ivanov for acquaintance". Document circulation without a mouse and the keypad becomes simply magnificent.

15

Re: Programming natural languages, move?

Hello, Ops, you wrote: Ops> It is all well for primitive tasks. And how about "create to me the program of document circulation for my firm"? As the person automating a heap of document circulations, authoritatively I declare - this phrase enough.

16

Re: Programming natural languages, move?

Hello, s_aa, you wrote: _> What for? "Register the document", "Find the document", "Transfer to Ivanov for acquaintance". Document circulation without a mouse and the keypad becomes simply magnificent. At first it is necessary to explain somehow these operations to the machine that is reduced to creation of system of document circulation.

17

Re: Programming natural languages, move?

Hello, Real 3L0, you wrote: R3> As the person automating a heap of document circulations, authoritatively I declare - this phrase enough. Even if firm - Gazprom?

18

Re: Programming natural languages, move?

Hello, Ops, you wrote: Ops> Even if firm - Gazprom? It is enough Lukoil and Rosatom?

19

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: IQ> If I can not fairly understand what tasks "programming in a natural language at all" can solve, since  the majority of programmers manage nasty to program and in the formal languages. Excuse, I can not simply be retained:

20

Re: Programming natural languages, move?

Ops> And whence it is known, what it generally can be reached? Here people cannot force to formulate normally customers of the requirement, and you about the machine. So by analogy to normal programming - with it somehow something reach. Programming process in a natural language it will be not so mandatory less long and painful, he simply allows to lower requirements to qualification of the programmer (but not the business analyst - actually will be programmed now by an analyst!) people in dialogue with the customer does not suffice either time, or patience, meticulousness and understanding that formalization yet did not reach 100 % - at the computer with all it problems are not present. Whether the new technology  (removing the programmer, but loading an analyst ) - a question, naturally, cardinal allows to save. At least, on time of dialogue of an analyst with the programmer, including through documentation writing/reading (and it is hours of two persons!) saving explicit.

21

Re: Programming natural languages, move?

Hello, Real 3L0, you wrote: R3> Hello, Ops, you wrote: Ops>> Even if firm - Gazprom? R3> it is enough Lukoil and Rosatom? I.e. programming in a natural language is reduced to the decision on implementation of SAP or something similar In that case the conceptual model such "a programming natural language" can be reduced to one syntactic unit - "implement!". At once the ancient roller https://www.youtube.com/watch?v=-uGdaKFboHo is recalled

22

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: IQ> I.e. programming in a natural language is reduced to the decision on implementation of SAP or something similar Re-read my example above. Still I will add that except "by default", the system should be able to do outputs. For example, if in system one printer silly to ask the user, on what printer to send to print.

23

Re: Programming natural languages, move?

Hello, Real 3L0, you wrote: R3> Hello, IQuerist, you wrote: IQ>> I.e. programming in a natural language is reduced to the decision on implementation of SAP or something similar R3> Re-read my example above. R3> Still I will add that except "by default", the system should be able to do outputs. Not "system" this business to do outputs. The system should offer decisions to the person, instead of suffer . R3> For example if in system one printer, silly to ask the user, on what printer to send to print. Really silly, the probability of that this printer is in other office for 1000 km, is scornfully small

24

Re: Programming natural languages, move?

Hello, IQuerist, you wrote: IQ> Not "system" this business to do outputs. The system should offer decisions to the person, instead of suffer . Or you are terribly far from the modern technologies, or "it is stupid ". In about what I wrote, "outputs" and "decisions" can be understood synonyms. A hint in this case - RDF, OWL. IQ> it is really silly, probability of that this printer is in other office for 1000 km, is scornfully small Read once again about outputs. A hint in this case - GPS.

25

Re: Programming natural languages, move?

Hello, Real 3L0, you wrote: R3> Hello, IQuerist, you wrote: IQ>> Not "system" this business to do outputs. The system should offer decisions to the person, instead of suffer . R3> Or you are terribly far from the modern technologies, or "it is stupid ". R3> In about what I wrote, "outputs" and "decisions" can be understood synonyms. I so understand the essential difference has an effect on experience "Decision", assumes "responsibility in difference from"outputs". IQ>> it is really silly, probability of that this printer is in other office for 1000 km, is scornfully small R3> Read once again about outputs. R3> the Hint in this case - GPS. ... You should learn about this world more so much