Summary of the invention
The embodiment of the present invention provides intelligent translation method and system in a kind of input method, carries out needing to carry out pattern switching frequently when different language exchanges, make input method use the problem of fluency variation to solve existing input method user.
For this reason, the embodiment of the present invention provides following technical scheme:
A kind of intelligent translation method in input method, comprising:
Input method instruction is called in reception, and receives the input information of user;
The language that the take over party obtaining described input information expects;
Judging that whether the language that receiving party expects is identical with the language of input information, if not, is then the language that take over party expects by described input information translation;
Information is inputted after output processing.
Preferably, the language that the take over party of the described input information of described acquisition expects comprises:
The relevant information of application and/or user according to calling input method input history, the language that the take over party obtaining described input information expects.
Preferably, the relevant information of the application of described input method comprise following any one or multiple: the sub-attribute information obtaining storage in apply property, application in advance, the geographical location information applied when calling input method.
Preferably, described basis calls the relevant information of the application of input method and/or user inputs history, and the language that the take over party obtaining described input information expects comprises:
The relevant information calling the application of input method by described and/or user input history, judge, the language that the take over party obtaining described input information expects with the judgement mapping table preset; Or
The relevant information calling the application of input method by described and/or user input history, according to the rule list preset, score value quantification is carried out to the language that the take over party of described input information expects, and according to the decision threshold preset, the score value quantized is judged, the language that the take over party obtaining described input information expects; Or
In advance based on the thought training judgment models returned, and the matching score of the feature and described judgment models that input history according to the relevant information of application and/or user of calling input method judges, obtains the language of take over party's expectation of described input information.
Preferably, described method also comprises:
Whether language receiving party expected is identical with the language of input information is divided into Three Estate: identical, possible identical, not identical, described score value≤the Second Threshold that may be all first threshold≤described quantification mutually, or the 3rd threshold value≤described matching score≤the 4th threshold value;
When the language that described receiving party expects is identical with the language possibility of described input information, then by first determining whether that needing described input information translation is the language that take over party expects with user interactions, or translation result and untranslated result are shown simultaneously so that user selects.
Preferably, described language comprises languages and dialect.
Intelligent translation system in a kind of input method, comprising:
Receiver module, calls input method instruction for receiving, and receives the input information of user;
Acquisition module, the language that the take over party for obtaining described input information expects;
Judge module, whether identical with the language of input information for judging the language that receiving party expects, if not, be then the language that take over party expects by described input information translation;
Output module, for inputting information after output processing.
Preferably, described acquisition module is specifically for inputting history according to the relevant information of application and/or user of calling input method, the language that the take over party obtaining described input information expects.
Preferably, described system also comprises:
Diversity module, is connected with judge module, and whether the language for receiving party being expected is identical with the language of input information is divided into Three Estate: identical, possible identical, not identical;
Interactive module, for when the language that described receiving party expects is identical with the language possibility of described input information, then by first determining whether that needing described input information translation is the language that take over party expects with user interactions, or translation result and untranslated result are shown simultaneously so that user selects.
Preferably, described language comprises languages and dialect.
Intelligent translation method and system in the input method that the embodiment of the present invention provides, first the method receives calls input method instruction, and receive the input information of user, then the language that the take over party obtaining described input information expects, then judge that the language that receiving party expects is translated described input information with whether identical the determining whether of language of input information, after final output processing, input information.Whether the language expected due to method automatic decision receiving party provided by the invention is identical with the language of input information, and automatically determine whether to translate described input information according to judged result, automatically can carry out pattern switching, make input method use fluency better.
Further, intelligent translation method and system in the input method that the embodiment of the present invention provides, whether the language receiving party expected of the method is identical with the language of input information is divided into Three Estate: identical, may be identical, not identical, when judged result is for determining whether to translate by user interactions time possible identical, can avoid like this being difficult in the scene determining whether translate input information, proportion such as, in current web page shared by Chinese and english close to time or pre-entered information in there is Chinese and english etc. simultaneously, to not wish that the input information translated carries out the problem translated, improve the accuracy rate whether to the judgement that input information is translated.
Further, described language comprises languages and dialect.When the language that receiving party expects is certain dialect, the relatives of such as Chinese user get used to using certain dialect, and user's current residing environment is when preferably making in English, can be the dialect that the relatives of user get used to by the English information translation that user inputs; Can provide result according to the demand of user's different levels, when user wishes warmer with the communication of good friend, the dialect kind can expected according to good friend is linked up, and to increase the cordial feeling between good friend, increases the experience satisfaction of client.
Embodiment
In order to the scheme making those skilled in the art person understand the embodiment of the present invention better, first brief introduction is carried out to intelligent translation method and system in input method provided by the invention, the language that the method is expected by the take over party inputting information described in automatic acquisition, then automatically determine whether to translate input information according to it, existing input method can be solved carry out needing to carry out pattern switching frequently when different language exchanges user, make input method use the problem of fluency variation.
Below in conjunction with drawings and embodiments, the embodiment of the present invention is described in further detail.
As shown in Figure 1, be a kind of process flow diagram of intelligent translation method in the input method that provides of the embodiment of the present invention, comprise the following steps:
Step S01, receives and calls input method instruction, and receive the input information of user.
In the present embodiment, the input information of described user can be by phonetic entry, as can be received the voice messaging of user by equipment such as microphones; Also can by the input of other modes such as button, as user can adopt phonetic or the input mode such as hand-written to carry out target character input on input method panel, which kind of input mode user can select to adopt as required, and the embodiment of the present invention does not limit.
In addition, in the input method that the embodiment of the present invention provides, intelligent translation method is all effective under all text event detection scenes, such as, in edit page in the chat message input of QQ chat software, in blog, in the editor of WORD document, in the Data Enter of Microsoft Excel or input the scenes such as contact person in local communication record under, when input method instruction is called in reception, receive the input information of user.
Step S02, the language that the take over party obtaining described input information expects.
In the present embodiment, the language that the take over party of the described input information of described acquisition expects comprises: the relevant information of application and/or user according to calling input method input history, the language that the take over party obtaining described input information expects.Particularly, the relevant information of the application of described input method comprise following any one or multiple: the sub-attribute information obtaining storage in apply property, application in advance, the geographical location information applied when calling input method.
Wherein, the take over party of described input information can be individual, application or website etc., such as when good friend is Australian, then its language expected is English, when using the application software of certain United States version, then its language expected is English, and when certain Russian website of login, then its language expected is Russian; Described apply property can refer to the type applied, as English website, Chinese website, social application software, short message service etc.; The sub-attribute information obtaining storage in described application in advance can refer in the social class application such as similar QQ, micro-letter, the good friend preset by historical information statistical study or user or the attribute of contact person, as some good friend uses Chinese, and part good friend is had to use English etc.; The described geographical location information applied when calling input method can refer to the geographic position at the mobile terminal place when certain application call input method, as being the area of certain language in official language and/or general purpose language, or the address having attribute that is that some mark according to user oneself or that obtain according to historical information statistical study, need to start interpretative function in certain western-style restaurant, certain embassy of state etc. as obtained user according to historical information analysis more, then this address is just with the attribute opening interpretative function, in this no limit.
Further, described language comprises languages and dialect.When the language that receiving party expects is certain dialect, the relatives of such as user get used to using certain dialect, and when the current residing environment of user preferably makes in English, can be the dialect that the relatives of user get used to by the English information translation that user inputs.In actual use, can also carry out classification to language, such as language comprises multiple languages, as Chinese, English, German, French, Japanese etc.; The corresponding multiple dialect of each languages, as Chinese can comprise northeast words, Sichuan words, Cantonese etc., like this, the language expected as receiving party may be certain languages, when also may be certain dialect of these languages, by first determining whether that needing described input information translation is the dialect kind that take over party expects with user interactions, or translation result and untranslated result can be shown simultaneously so that user selects.Like this, can provide result according to the demand of user's different levels, when user wishes warmer with the communication of good friend, the dialect kind choice for use dialect can expected according to good friend is linked up, to increase the cordial feeling between good friend, increase the experience satisfaction of client.Certainly, can also be the another kind of dialect of identical languages by certain dialect translation, or be certain dialect of another languages by certain dialect translation, not limit at this.
It should be noted that, the take over party of described input information can be individual, application or website etc.Individually, its language expected can input history to determine according to the relevant information of application and/or user calling input method, such as mother tongue; Its dialect expected is its on-site dialect in permanent residence/native place is excellent, such as Xinjiang dialect: the show off spare time passes, show off droops subrepresentation chat, " the chatting " in " chatting " northeast in " chatting idly " namely in Beijing native language, Sichuan, recklessly, soup represents muddleheaded, at sixes and sevens, and my face of the smell of urine represents to my embarrassment etc.; Again for Chongqing Dialect: tired make represent very dirty, frightened expression of grabbing is made a fuss, pin enters male cousin and shows that pin has been rivered bend, eat vast expression to have a meal, when take over party gets used to using dialect, and user is current when being not easy to/can not using dialect, and such as duration of work relatives and friends send news in brief, the mandarin that user inputs is translated as dialect and the other side carries out linking up the effect that can promote to link up, the cordial feeling of increase both sides.With regard to application or website, its dialect kind expected is excellent with Official Languages, such as, when with phonitic entry method input information, people usually get used to local dialect, such as Wenzhou dialect: eat a day light representations and have breakfast, and eat to represent a day daytime and have lunch, with not represent cannot, pleasure is aobvious represents very suitable; Again for Inner Mongol dialect: needless to say will represent and be disinclined to do, vigorously the expression of youngster loses no time/hurries up, give a tongue-lashing meaning with thumb down, do the wash a lattice watt expression to wash away, when dialect being inputted above-mentioned application or website, be often not easy to apply or website understand user want express true intention, can be now Official Languages by these dialect translations, improve understandability, discrimination etc., to improve Consumer's Experience effect.
Particularly, described basis calls the relevant information of the application of input method and/or user inputs history, the language that the take over party obtaining described input information expects comprises: the relevant information calling the application of input method by described and/or user input history, judge with the judgement mapping table preset, the language that the take over party obtaining described input information expects; Or call the relevant information of application of input method described in passing through and/or user inputs history, according to the rule list preset, score value quantification is carried out to the language that the take over party of described input information expects, and according to the decision threshold preset, the score value quantized is judged, the language that the take over party obtaining described input information expects; Or in advance based on the thought training judgment models returned, and the matching score of the feature and described judgment models that input history according to the relevant information of application and/or user of calling input method judges, obtains the language of take over party's expectation of described input information.Certain above-mentioned rule list and judgment models also can be combined, such as can there are rule list and judgment models two kinds of judgment modes simultaneously, adopt rule list to judge according to the threshold value preset when data volume deficiency, and when data accumulation some, adopt judgment models to judge according to corresponding default threshold value.
Wherein, judge that mapping table is generally input information and the mapping table of corresponding language, if its attribute of application calling input method is English website, now judge that the Feature Mapping of this input information in mapping table is to English; The apply property and for example calling input method is QQ, and it is applied sub-attribute information for certain good friend and uses Hefei to talk about, then mapping table is mapped to Hefei words etc.; Rule list is to judging that mapping table quantizes, if apply property is English website, quantizing score value is 0.9 point, total score 1, in addition, the output of rule list is the score value quantized, and is convenient to consider features such as multiple relevant informations, such as be weighted summation etc. by the score value of the quantification to various relevant information feature, the language that obtaining information take over party expects; The relevant information being input as the application of calling input method of judgment models and/or user input the feature of history, export as each language matching score, such as support vector machine (SupportVectorMachine, SVM), neural network (NeuralNetworks, NN) models etc., are not enumerating.
Step S03, judges that whether the language that receiving party expects is identical with the language of input information, if not, is then the language that take over party expects by described input information translation.
In actual applications, whether the language that the receiving party that can obtain according to step S02 expects is identical with the language of input information, judges whether to start interpretative function.
Further, described method also comprises: whether language receiving party expected is identical with the language of input information is divided into Three Estate: identical, may be identical, not identical, described score value≤the Second Threshold that may be all first threshold≤described quantification mutually, or the 3rd threshold value≤described matching score≤the 4th threshold value, such as, described first threshold can be 0.4, described Second Threshold can be 0.7, described 3rd threshold value can be 0.4, described 4th threshold value can be 0.7, be only exemplary with upper threshold value, limitation of the present invention can not be interpreted as, the setting of each threshold value is specifically depending on practical effect, when the language that described receiving party expects is identical with the language possibility of described input information, then by first determining whether that needing described input information translation is the language that take over party expects with user interactions, or translation result and untranslated result are shown simultaneously so that user selects.Certainly, whether the language that can receiving party be expected according to the actual requirements is identical with the language of input information is divided into more grade, is not enumerating.Like this, when judged result is possible identical, determine whether to translate by user interactions, can avoid being difficult in the scene determining whether translate input information, proportion such as, in current web page shared by Chinese and english close to time or pre-entered information in there is Chinese and english etc. simultaneously, to not wish that the input information translated carries out the problem translated, improve the accuracy rate whether to the judgement that input information is translated.Concrete, according to practical situations, in order to reduce erroneous judgement, can judgment threshold be set stricter, or when model training by stricter for the data mark opening interpretative function.When carrying out multistage judgement, Three Estate as can be whether identical according to language: identical, may be identical, not identical, whether will open interpretative function to be divided into " unlatching interpretative function ", " may need to open interpretative function ", three kinds of situations such as " do not open interpretative function ", only the obvious situation of intention is judged to be " unlatching interpretative function " or " not opening interpretative function ", and for the situation that may occur judging by accident, elder generation and user interactions can be passed through, thus determine whether to need to open, or so that translation result and untranslated result are shown in candidate bar simultaneously, then determination result is selected according to user.Can avoiding like this being difficult in the scene determining whether translate input information, by not wishing that the input information translated carries out the problem translated, improving the accuracy rate whether input information translated.
Certainly, when step S02 also obtains the dialect that receiving party expects, can be directly the dialect that receiving party expects by described input information translation; Also can be first the languages that take over party expects by described input information translation according to the setting of user, then according to user's request, translation result can be translated into again the dialect that take over party expects, in this no limit.
It should be noted that, this interpretative function can be embedded in input method, can certainly be to call external translation software to translate, and does not make particular determination at this.
Step S04, inputs information after output processing.
In the present embodiment, when the language of the information of input is identical with the language that debit waiting expects, the identification decoded result of user's input information in input method can be provided, identifying as adopted the technology such as existing speech recognition, coded string identification; When the language of the information of input is different from the language that debit waiting expects, export the result after translation.
Intelligent translation method in the input method that the embodiment of the present invention provides, the method calls input method instruction by reception, and receive the input information of user, then the language that the take over party obtaining described input information expects, then judge that the language that receiving party expects is translated described input information with whether identical the determining whether of language of input information, after final output processing, input information.Due to the language that the take over party inputting information described in energy automatic acquisition expects, then automatically determine whether to translate input information according to it, existing input method can be solved carry out needing to carry out pattern switching frequently when different language exchanges user, make input method use the problem of fluency variation.
Correspondingly, the present invention also embodiment provides intelligent translation system in a kind of input method, as shown in Figure 2, is a kind of structural representation of this system.
In this embodiment, described system comprises:
Receiver module 201, calls input method instruction for receiving, and receives the input information of user;
Acquisition module 202, the language that the take over party for obtaining described input information expects;
Judge module 203, whether identical with the language of input information for judging the language that receiving party expects, if not, be then the language that take over party expects by described input information translation;
Output module 204, for inputting information after output processing.
In the present embodiment, described acquisition module 202 is specifically for inputting history according to the relevant information of application and/or user of calling input method, the language that the take over party obtaining described input information expects.
In order to improve the accuracy whether starting interpretative function, described system also comprises diversity module 305 and interactive module 306, and as shown in Figure 3, in actual applications, described system also comprises:
Diversity module 305, be connected with judge module, whether the language for receiving party being expected is identical with the language of input information is divided into Three Estate: identical, possible identical, not identical, described score value≤the Second Threshold that may be all first threshold≤described quantification mutually, or the 3rd threshold value≤described matching score≤the 4th threshold value;
Interactive module 306, for when the language that described receiving party expects is identical with the language possibility of described input information, then by first determining whether that needing described input information translation is the language that take over party expects with user interactions, or translation result and untranslated result are shown simultaneously so that user selects.
Further, described language comprises languages and dialect.To meet the dialect translation demand of user.
Certainly, this system can further include memory module (not shown), the relevant informations such as history are inputted, as the usual dialect of certain relatives and friends for preserving speech encoding, input information, model parameter, the relevant information of application of input method and/or user.Like this, computing machine is carried out to the input information of user automatically process to facilitate, and store the relevant informations such as translation result.
Intelligent translation system in the input method that the embodiment of the present invention provides, first this system is received by receiver module 201 and calls input method instruction, and receive the input information of user, then the language that the take over party obtaining described input information by acquisition module 202 expects, then judge that whether the language that receiving party expects is identical with the language of input information by judge module 203, if not, be then the language that take over party expects by described input information translation, input information eventually through after output module 204 output processing.Due to acquisition module 202 can automatic acquisition take over party expect language, then translated the need of to input information by judge module 203 automatic decision, whether native system is translated input information without the need to artificial selection, the language of the result that output module 204 exports is the language that take over party expects, makes input method use fluency better.
Each embodiment in this instructions all adopts the mode of going forward one by one to describe, between each embodiment identical similar part mutually see, what each embodiment stressed is the difference with other embodiments.Especially, for system embodiment, because it is substantially similar to embodiment of the method, so describe fairly simple, relevant part illustrates see the part of embodiment of the method.System embodiment described above is only schematic, the wherein said unit illustrated as separating component or can may not be and physically separates, parts as unit display can be or may not be physical location, namely can be positioned at a place, or also can be distributed in multiple network element.Some or all of module wherein can be selected according to the actual needs to realize the object of the present embodiment scheme.Those of ordinary skill in the art, when not paying creative work, are namely appreciated that and implement.
Being described in detail the embodiment of the present invention above, applying embodiment herein to invention has been elaboration, the explanation of above embodiment just understands method of the present invention, system and earphone for helping; Meanwhile, for one of ordinary skill in the art, according to thought of the present invention, all will change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.