CN105335415A - Search method based on input prediction, and input method system - Google Patents
Search method based on input prediction, and input method system Download PDFInfo
- Publication number
- CN105335415A CN105335415A CN201410379907.3A CN201410379907A CN105335415A CN 105335415 A CN105335415 A CN 105335415A CN 201410379907 A CN201410379907 A CN 201410379907A CN 105335415 A CN105335415 A CN 105335415A
- Authority
- CN
- China
- Prior art keywords
- candidate
- sequence
- character string
- current input
- input
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention provides a search method based on input prediction, and an input method system, and is used for solving the problems of complexity in a traditional search operation and low efficiency. The method comprises the following steps: receiving current input information input by a user, wherein the current input information comprises a current input character string and time interval information; on the basis of the time interval information, segmenting the current input character string in real time, and carrying out conversion to obtain a candidate sequence set corresponding to the current input character string; and independently searching each candidate sequence in the candidate sequence set, obtaining a search result independently corresponding to each candidate sequence, and carrying out feedback on the basis of the candidate sequence set and the corresponding search result. In an input process, a segmentation result can be quickly and conveniently obtained according to the real-time segmentation of the time interval information, an input matching degree is improved, and the desired search result can be obtained without carrying out input or search in a search box of the search engine. The search method has the advantages of being convenient and quick in operation and improves search efficiency.
Description
Technical field
The present invention relates to input method technique field, particularly relate to a kind of searching method based on input prediction and a kind of input method system.
Background technology
When user searches for information in a network, normally open any browser, main browser page search box or enter search engine in a browser searched page in carry out searching for.
In search box or searched page, use input method to knock a string list entries of input through keyboard, input method provides several candidate word, and user constantly selects the candidate word that satisfies the demands is shielded, until expressed intact user can want the keyword searched for.Then user clicks " search " button transmission searching request and obtains Webpage searching result.
Prior art needs user constantly to screen in each candidate item obtained according to list entries, and in the target candidate obtained after screen, then triggering searches order obtains corresponding Search Results, and complex operation, efficiency is lower.
Summary of the invention
Embodiment of the present invention technical matters to be solved is to provide a kind of searching method based on input prediction, loaded down with trivial details to solve existing search operation, the problem that efficiency is lower.
Accordingly, the embodiment of the present invention additionally provides a kind of input method system, in order to ensure the implementation and application of said method.
In order to solve the problem, the invention discloses a kind of searching method based on input prediction, comprising: the current input information receiving user's input, wherein, described current input information comprises current input character string and time interval information; In real time cutting is carried out to described current input character string according to described time interval information, and be converted to collection of candidate sequences corresponding to described current input character string; Each candidate sequence in described collection of candidate sequences is searched for respectively, obtains each self-corresponding Search Results of each candidate sequence, and feed back according to described collection of candidate sequences and corresponding Search Results.
Optionally, described time interval information comprises the dead time of every two adjacent characters in described current input character string; Describedly in real time according to described time interval information described current input character string is carried out to cutting and is converted to collection of candidate sequences, comprise: when detecting that the dead time produced in real time in input process meets pre-conditioned, the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding; Each cutting sequence is changed successively and combined, obtains the collection of candidate sequences that described current input character string is corresponding.
Optionally, described current input information also comprises user account information; Describedly in real time cutting carried out to described current input character string according to described time interval information and before being converted to candidate sequence, also comprise: the average dead time of searching respective user according to described user account information; Adopt the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtain the normalization dwell interval produced in real time.
Optionally, also comprise: by the normalization dwell interval produced in real time in described current input character string, compare with pause threshold value successively; When the normalization dwell interval produced in real time in input process is greater than described pause threshold value, determine that the dead time of described real-time generation meets pre-conditioned.
Optionally, described method also comprises the determining step of the average dead time of user: determine user according to described user account information, and collects the history input information of described user; The time of pausing after inputting single character according to user described in described history input Information Statistics, as the average dead time of described user.
Optionally, each cutting sequence changed successively and combined, obtaining the collection of candidate sequences that described current input character string is corresponding, comprising: at least one the candidate Chinese character string searching each cutting sequence from preset index respectively; The candidate Chinese character string of each cutting sequence is combined, obtains each Chinese character sequence that described input of character string is corresponding respectively; In language model, obtain the probability of each candidate Chinese character sequence respectively, and choose described each Chinese character Sequence composition collection of candidate sequences respectively according to described probability.
Optionally, describedly choose described each Chinese character Sequence composition collection of candidate sequences respectively according to described probability, comprise: each Chinese character sequence is sorted according to described probability is descending, and each Chinese character sequence coming top N is formed collection of candidate sequences as candidate sequence, wherein N is positive integer.
Optionally, describedly choose described each Chinese character Sequence composition collection of candidate sequences according to described probability, comprising: each Chinese character sequence is sorted according to described probability is descending, choose the Chinese character sequence of maximum probability as keyword; From the index database of search engine, inquiry and described keyword have the search word of correlativity, and sort according to characteristic of correspondence value to the search word inquired; Choose the search word coming front M position and form described collection of candidate sequences respectively as candidate sequence, wherein M is positive integer.
Optionally, described each candidate sequence in described collection of candidate sequences to be searched for respectively, obtain each self-corresponding Search Results of each candidate sequence, comprising: in a search engine each candidate sequence in described collection of candidate sequences is searched for successively; For the Search Results of each candidate sequence, screen the Search Results coming front X position of described search engine feedback, wherein X is positive integer.
Optionally, described method also comprises: show respectively described collection of candidate sequences and corresponding Search Results, be specially: show successively the candidate sequence in described collection of candidate sequences; Obtain and be illustrated in each Search Results corresponding to the first candidate sequence and show.
Optionally, also comprise: receive and instruction is chosen to described candidate sequence, determine the candidate sequence chosen; Each Search Results that the candidate sequence chosen described in acquisition is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is shown.
Accordingly, the invention also discloses a kind of input method system, comprising: receiver module, for receiving the current input information of user's input, wherein, described current input information comprises current input character string and time interval information; Modular converter, for carrying out cutting to described current input character string in real time according to described time interval information and being converted to collection of candidate sequences corresponding to described current input character string; Search module, for searching for respectively each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence; Feedback module, for feeding back according to described collection of candidate sequences and corresponding Search Results.
Optionally, described time interval information comprises the dead time of every two adjacent characters in described current input character string; Described modular converter, comprise: cutting submodule, for when detecting that the dead time produced in real time in input process meets pre-conditioned, the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding; Transform subblock, for changing successively each cutting sequence and combine, obtains the collection of candidate sequences that described current input character string is corresponding.
Optionally, described current input information also comprises user account information; Described modular converter, also comprises: normalization submodule, for searching the average dead time of respective user according to described user account information; Adopt the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtain the normalization dwell interval produced in real time.
Optionally, described cutting submodule, the normalization dwell interval also for producing in real time in described current input character string, compares with pause threshold value successively; When the normalization dwell interval produced in real time in input process is greater than described pause threshold value, determine that the dead time of described real-time generation meets pre-conditioned.
Optionally, described search module, comprising: search submodule, for searching for successively each candidate sequence in described collection of candidate sequences in a search engine; Screening submodule, for the Search Results for each candidate sequence, screen the Search Results coming front X position of described search engine feedback, wherein X is positive integer.
Optionally, described system also comprises: display module, for showing respectively described collection of candidate sequences and corresponding Search Results; Wherein, described display module, comprising: first shows submodule, for showing successively the candidate sequence in described collection of candidate sequences; Second shows submodule, is illustrated in each Search Results corresponding to the first candidate sequence for obtaining and shows.
Optionally, described system also comprises: indicating module, choosing instruction, determining the candidate sequence chosen for receiving to described candidate sequence; Then described second shows submodule, and also for each Search Results that the candidate sequence chosen described in obtaining is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is shown.
Compared with prior art, the embodiment of the present invention comprises following advantage:
Receive the current input information comprising current input character string and time interval information of user's input, input prediction is carried out by current input information, wherein time interval information characterizes the dead time of current input character string in input process, therefore in the input process of user, carrying out cutting to current input character string in real time according to time interval information can be accurate, cutting word occurrence efficiently or character string corresponding to word, improve Input matching degree, thus be converted to collection of candidate sequences accurately, adopt each candidate sequence in this collection of candidate sequences to carry out search respectively again and obtain Search Results, and feed back according to described collection of candidate sequences and corresponding Search Results, without the need to screen on word directly being obtained by input method the Search Results wanted in the search box of search engine, easy and simple to handle, fast, improve search efficiency.
Accompanying drawing explanation
Fig. 1 is the flow chart of steps of a kind of searching method embodiment based on input prediction of the present invention;
Fig. 2 is the flow chart of steps of another kind of the present invention based on the searching method embodiment of input prediction;
Fig. 3 is the flow chart of steps of a kind of searching method embodiment based on input prediction of the present invention;
Fig. 4 is the flow chart of steps of pre-conditioned detection in the embodiment of the present invention two;
Fig. 5 is candidate Chinese character string combination schematic diagram in the embodiment of the present invention three;
Fig. 6 is the first displaying schematic diagram of input method client in the embodiment of the present invention three;
Fig. 7 is that in the embodiment of the present invention three, input method client the second shows schematic diagram;
Fig. 8 is the structured flowchart of a kind of input method system embodiment of the present invention;
Fig. 9 is the structured flowchart of the first embodiment of input method system in the embodiment of the present invention;
Figure 10 is the structured flowchart of the second embodiment of input method system in the embodiment of the present invention.
Embodiment
For enabling above-mentioned purpose of the present invention, feature and advantage become apparent more, and below in conjunction with the drawings and specific embodiments, the present invention is further detailed explanation.
One of core idea of the embodiment of the present invention is, proposes a kind of searching method based on input prediction, loaded down with trivial details to solve existing search operation, the problem that efficiency is lower.Receive the current input information comprising current input character string and time interval information of user's input, input prediction is carried out by current input information, wherein time interval information characterizes the dead time of current input character string in input process, therefore in the input process of user, carrying out cutting to described current input character string in real time according to time interval information can be accurate, cutting word occurrence efficiently or character string corresponding to word, improve Input matching degree, thus be converted to collection of candidate sequences accurately, adopt each candidate sequence in this collection of candidate sequences to carry out search respectively again and obtain Search Results, and described collection of candidate sequences and corresponding Search Results are directly fed back to described input method client, directly can obtain by input method the Search Results wanted without the need to inputted search in the search box of search engine, easy and simple to handle, fast, improve search efficiency.
Embodiment one
With reference to Fig. 1, show the flow chart of steps of a kind of searching method embodiment based on input prediction of the present invention, specifically can comprise the steps:
Step 102, server receives the current input information that input method client is uploaded.
The embodiment of the present invention carries out searching for improve search efficiency according to input of character string in the input process of input method.Namely the character that input method client is receiving user's input is formed in the process of input of character string, in real time current input information is uploaded to server.Wherein, described current input information comprises current input character string and time interval information, described time interval information for characterizing interval input time in current input character string between character, i.e. the dead time of current input character string in input process.
Wherein, input method client carries out the upload operation of current input information in real time, therefore after user often inputs a character, all the time interval information of this character and this character and previous character can be uploaded, can be now that increment is uploaded, also can with input before do not determine that other character strings of upper screen option merge and upload, the embodiment of the present invention does not limit this.
The current input information that server can be uploaded based on input method client carries out follow-up input prediction and search.
Step 104, server carries out cutting to described current input character string in real time according to described time interval information and is converted to collection of candidate sequences corresponding to described current input character string.
Pause when can determine that character inputs by time interval information, usual user is when inputting series of characters string, when inputting each character of a word or word continuously, a shorter pause can be there is in each intercharacter, and at a word or complete character string corresponding to word after input completes, usually a longer pause can be there is after character string, therefore by the length of intercharacter dead time can real-time judgment current be just an input word or the corresponding intermediate character of word or inputted a complete complete character string, thus the cutting realized in real time to input of character string.
Server obtains each intercharacter dead time in current input character string from time interval information, when determining to be input to certain character according to the dead time, that the input of single character completes or the input of a word or complete character string corresponding to word completes, thus after judging to have inputted character string corresponding to words, real-time cutting is carried out at this character place, then detect, until the whole cutting of current input character string is complete based on to successive character in described current input character string.Then the character string after cutting is changed in real time, obtain each candidate sequence that current input character string is corresponding, thus form collection of candidate sequences.
Step 106, server is searched for respectively to each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence, and described collection of candidate sequences and corresponding Search Results are fed back to described input method client.
After going out candidate sequence by real-time cutting, conversion estimation, from collection of candidate sequences, take out each candidate sequence in real time search for successively, determine the Search Results of each candidate sequence, also can the Search Results of only retrieving portion candidate sequence, follow-uply again the candidate sequence of other parts to be searched for according to demand.Then collection of candidate sequences and corresponding Search Results are fed back to input method client.
Wherein, when feeding back Search Results, once the Search Results of each candidate sequence all can be fed back, and input method client is after receiving collection of candidate sequences and Search Results corresponding to each candidate sequence, can show respectively each candidate sequence and Search Results, as shown each candidate sequence in collection of candidate sequences successively, simultaneously exposition or Search Results that all candidate sequence is corresponding.
Also only can give tacit consent to first and feed back the corresponding Search Results of the candidate sequence ranked the first, show fast for the first time for user, and utilize user to check, or carry out keyboard operation when the first sequence being moved to the time slot of non-the first sequence, transmit the Search Results that non-the first sequence pair answers separately to show to input method client, thus ensure that the data that Search Results is corresponding can be transmitted fast, show, check for user.
The present embodiment in input method after input of character string namely by determining candidate sequence to the prediction of input of character string, and then search candidate sequence Search Results show, be convenient to search for fast.
Above provide by the mutual realization of input method client and the server search embodiment based on input prediction, in actual treatment, independently can also be performed the prediction of input by input method client, only adopt server to search for, specifically comprise following operation steps as shown in Figure 2:
Step 202, input method client receives the current input information of active user's input.
Step 204, input method client carries out cutting to described current input character string in real time according to described time interval information, and is converted to collection of candidate sequences corresponding to described current input character string.
Step 206, gives server by above-mentioned for the candidate sequence in described collection of candidate sequences.
Step 208, server is searched for respectively to each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence, and described collection of candidate sequences and corresponding Search Results are fed back to described input method client.
The current input character string of input method client user in real input in user's input process and time interval information, thus cutting can be carried out to inputting dead time longer two intercharacters in current input character string in real time according to time interval information, such as, real-time calculating or obtain the average input interval of current input character string in advance from server, then determine at two larger characters of input dead time according to average input interval, thus determine the position of intercharacter cutting, realize according to the cutting of time to current input character string, further collection of candidate sequences is converted to the character string after cutting again.
Again collection of candidate sequences is uploaded to server, the part or all of candidate sequence in collection of candidate sequences now can be uploaded according to network condition, thus adopt server candidate sequence is searched for, search step and above-mentioned steps 106 basically identical, therefore repeat no more.
To sum up, the present embodiment can be determined in server or input method client, perform the operation steps according to time real-time cutting character string according to network condition, thus real-time estimate goes out the candidate sequence that user may input, and then quick-searching is carried out to candidate sequence, raise the efficiency.
Embodiment two
On the basis of above-described embodiment, the present embodiment is discussed further and is predicted and the method for searching for based on prediction input.
With reference to Fig. 3, show the flow chart of steps of a kind of searching method embodiment based on input prediction of the present invention, specifically can comprise the steps:
Step 302, server receives the current input information that input method client is uploaded.
Wherein, described current input information comprises current input character string, time interval information and user account information, user account information is used for identifying user, different users has the information such as different input habits, historical data, the embodiment of the present invention can be distinguished different users by user account information, to dope the candidate sequence meeting this user view most, obtain Search Results accurately and recommend.
Step 304, server is when detecting that the dead time produced in real time in input process meets pre-conditioned, and the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding.
Server is after receiving the current input information that input method client uploads, real-time cutting can be carried out to current input character string according to time interval information, wherein, time interval information comprises the dead time produced in real time in input process, this dead time comprises dead time of two adjacent characters produced in real time along with user's current input character string described in input process, namely in time interval information, real time record user inputs the dead time between character to input character late successively, if therefore current input character string comprises n character, then time interval information comprises the dead time of (n-1) individual adjacent character.
How the embodiment of the present invention pre-conditionedly carries out cutting to character for determining if being provided with, in user's input process, when detecting that the dead time of two adjacent characters produced in real time meets pre-conditioned, just can between these two adjacent characters, namely current input position place carries out cutting, thus judge whether the dead time produced in real time in input process meets pre-conditioned successively, just can determine how cutting is carried out to current input character string, obtain each cutting sequence that input of character string is corresponding.Certainly this current input of character string until correspondence candidate item on shield and also may not need cutting, the embodiment of the present invention is not construed as limiting this.
Wherein, pause threshold value can be set the dead time produced in real time in input process is evaluated, when the dead time produced in real time in input process is greater than this pause threshold value, think that it meets pre-conditioned, otherwise discontented pre-conditioned.
But because the input habit of different user is different, as old man's typing speed is comparatively slow, the average dead time inputting each character is long, and the average dead time that the skilled young man of computation inputs each character will be much short.Therefore when arranging pause threshold value, can arrange different pause threshold values for different users, the pause threshold value obtaining its correspondence according to user account information compares.
Unified pause threshold value can certainly be set for all users, but now will eliminate the individual difference impact of user's input habit, specifically comprise the steps: as shown in Figure 4
Step 402, searches the average dead time of respective user according to described user account information.
In order to eliminate the impact of the individual difference of user's input habit, first the average dead time of user is searched by user account information, wherein, the described average dead time refers to the dead time of user when continuous input character, as the dead time between two adjacent characters that user inputs a word or word continuously.
Wherein, the average dead time of user is determined as follows: determine user according to described user account information, and collects the history input information of described user; The time of pausing after inputting single character according to user described in described history input Information Statistics, as the average dead time of described user.
The embodiment of the present invention is distinguished the input habit without user according to user's history input data, first corresponding user is determined according to described user account information, then the history input information of each user is collected, as the character string of input can be comprised, the upper screen option that this character string is corresponding, the time interval of adjacent two characters in character string, the input time etc. of this section of character string, thus the time of pausing after inputting monocase according to this user of history input Information Statistics of each user respectively, as the average dead time of described user, " user-average input time " mapping table can also be set up at server end, for storing the average dead time of each user.Then the average dead time of described user account information respective user can be searched from " user-average input time " mapping table.
Certainly, in other embodiments, can also by identification informations such as the IP addresses according to certain computing machine, by server, the corresponding content of " user-average input time " mapping table relating to this computing machine is issued to this computing machine, makes input method client carry out real-time cutting according to the average dead time of respective user to current input character string.
Step 404, adopts the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtains the normalization dwell interval produced in real time.
Then the average dead time of this user is adopted to be normalized the dead time produced in real time in input process, by dead time of every two adjacent characters respectively divided by the average dead time of this user, eliminate the individual difference between different user, obtain the normalization dwell interval produced in real time.
Step 406, detects the normalization dwell interval produced in real time in described current input character string successively and whether is greater than pause threshold value.
By the normalization dwell interval produced in real time in described current input character string, compare with pause threshold value successively, detect the normalization dwell interval produced in real time and whether be greater than described pause threshold value.
If so, the normalization dwell interval namely produced in real time is greater than described pause threshold value, performs step 408; If not, then return step 406 to continue to detect, until the normalization dwell interval of all adjacent characters of described current input character string all completes detection, as completed input etc.
Step 408, the dead time of described real-time generation meets pre-conditioned.
When the normalization dwell interval of any two adjacent characters is greater than described pause threshold value, determine that the dead time of described adjacent character meets pre-conditioned, follow-uply can carry out cutting between these two adjacent characters.
By the steps flow chart of above-mentioned Fig. 4, the individual difference impact of user's input habit can be eliminated, thus whether the dead time produced in real time meets to adopt unified pause threshold value to determine pre-conditioned, and meeting pre-conditioned rear execution step 306.
Step 306, server is changed successively each cutting sequence and combines, and obtains the collection of candidate sequences that described current input character string is corresponding.
Input of character string may be split out multiple cutting sequence, one or more candidate Chinese character string is converted to respectively to each cutting sequence, multiple cutting sequence is there is at input of character string, can combine not homotactic candidate Chinese character string, obtain the collection of candidate sequences that described input of character string is corresponding.Certain input of character string itself also may it can be used as a cutting sequence without the need to cutting, now directly can be converted to the direct alternatively Sequence composition collection of candidate sequences of one or more candidate Chinese character string.
In the present invention's embodiment, each cutting sequence is changed successively and combined, obtains the collection of candidate sequences that described input of character string is corresponding, comprising: at least one the candidate Chinese character string searching each cutting sequence from preset index respectively; The candidate Chinese character string of each cutting sequence is combined, obtains each Chinese character sequence that described input of character string is corresponding respectively; In language model, obtain the probability of described each Chinese character sequence, and choose described each Chinese character Sequence composition collection of candidate sequences according to described probability.
In the embodiment of the present invention, preset index can be " pinyin-Hanzi " index, the character string of the phonetic inputted by user corresponding conversion can obtain Chinese character, and wherein " pinyin-Hanzi " index is that system is passed through Data Collections such as user search data, input method data automatic marking phonetics, analyzed rear foundation.Also be provided with language model, language model is the probability model set up for certain language, wherein, the probable value of correct word sequence is greater than the probable value of erroneous words sequence, system uses search data train language model, the follow-up probability that can be calculated a Chinese character sequence search word by language model.
Therefore, at least one candidate Chinese character string of each cutting sequence is searched respectively from preset " pinyin-Hanzi " index, then each candidate Chinese character string of all cutting sequences is combined, each Chinese character sequence that the input of character string after combining is corresponding can be obtained, such as current input character string is split out 2 cutting sequences, first cutting sequence is converted to 3 candidate Chinese character strings, and second cutting sequence is converted to 4 candidate Chinese character strings, then can be combined into 12 Chinese character sequences.
Then in language model, obtain the probability of described each Chinese character sequence, and choose described each Chinese character Sequence composition collection of candidate sequences according to described probability.
Wherein, describedly choose described each Chinese character Sequence composition collection of candidate sequences according to described probability, at least comprise following two kinds of modes:
(1) directly candidate sequence is chosen by sequence
Can sort according to described probability is descending to each Chinese character sequence, then therefrom extract each Chinese character sequence coming top N, by this N number of Chinese character sequence alternatively Sequence composition collection of candidate sequences, wherein N is positive integer.
(2) by meeting most the Chinese character sequence determination candidate sequence of user view
The Chinese character sequence of maximum probability meets user most to input intention, therefore can sort according to described probability is descending to each Chinese character sequence, chooses the Chinese character sequence of maximum probability as keyword.
Then search and meet user most and input related term corresponding to the keyword of intention, namely from the index database of search engine, inquiry and described keyword have the search word of correlativity, then the characteristic of correspondence value inquiring search word is obtained, wherein eigenwert can according to requirements set, as the correlativity etc. of user search number of times, keyword and search word, search word will be inquired again sort according to characteristic of correspondence value, choose the search word coming front M position and form described collection of candidate sequences respectively as candidate sequence, wherein M is positive integer.
Inputting the keyword predicted query search word of intention by meeting most user, can just predict that before user does not complete input user wants the content inputted.Such as, the original input of character string of user is " liude ", by being converted to keyword for " Liu De ", doping user want the content of searching for be " Liu Dehua ", " Liu De China film ", " concert of Liu De China " etc. by keyword " Liu De ".Thus the keyword query inputting intention by meeting most user has the search word of correlativity, user only needs importation character string, and user just can be helped to be described search intention, obtains inputting the input meeting oneself intention, can user operation be saved, promote input efficiency.
Step 308, server is searched for successively to each candidate sequence in described collection of candidate sequences in a search engine.
Step 310, server, for the Search Results of each candidate sequence, screens the Search Results coming front X position of described search engine feedback.
Then search engine is adopted to search for successively to each candidate sequence in collection of candidate sequences, query webpage index, summary etc. obtain each Search Results corresponding to each candidate sequence, each Search Results is sorted, can directly according to the sort result of search engine feedback, obtain the Search Results of front X position wherein, wherein X is positive integer.
Step 312, input method client is shown successively to the candidate sequence in described collection of candidate sequences.
Step 314, input method client obtains and is illustrated in each Search Results corresponding to the first candidate sequence and shows.
Described input method client, after receiving collection of candidate sequences and Search Results, can be shown as Search Hints the candidate sequence in collection of candidate sequences successively successively.Wherein, the displaying order of candidate sequence can be defined as the order of candidate sequence according to it, if as the candidate sequence directly chosen by sequence, then according to the size clooating sequence of its probability as displaying order, if by the candidate Chinese character sequence determination candidate sequence meeting most user view, then can according to the displaying order of the clooating sequence of its eigenwert as Search Hints.
Meanwhile, for being illustrated in the first candidate sequence, each Search Results that can also obtain its correspondence is shown.As each candidate sequence is illustrated in left side, and Search Results corresponding to the first candidate sequence is shown in the Search Results display area on right side.
In the present invention's embodiment, described input method client receives chooses instruction to described candidate sequence, is triggered, determine the candidate sequence chosen by modes such as directionkeys; Each Search Results that the candidate sequence chosen described in acquisition is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is synchronously shown in the Search Results display area on right side.
After input method client illustrates the Search Results of candidate sequence and the first candidate sequence, user may want the Search Results checking other candidate sequence, then user can by clicking, mouse moves, the modes such as sliding mouse roller send chooses instruction, input method client receives and chooses instruction to described candidate sequence, the corresponding candidate sequence chosen of instruction is chosen described in determining, then each Search Results that the candidate sequence chosen described in obtaining is corresponding, synchronous displaying after each Search Results that the candidate sequence chosen described in employing is corresponding carries out replacement to the Search Results of the first candidate sequence.
To sum up, the embodiment of the present invention is distinguished different users by user account information, to dope the candidate sequence meeting this user view most, obtains Search Results accurately and recommends.
Embodiment three
On the basis of above-described embodiment, the present embodiment adopts the example in concrete enforcement to discuss the method for carrying out searching for recommendation based on input prediction further.
Suppose that user wants search " Site of Qing Hua Yuan dream with glorious ", then carry out search recommendation step process in input prediction as follows:
User is inputted by input method, and current input character string is uploaded to server, and when character string being inputed to " qinghuayuan ", based on the time interval information uploaded in real time, server determines that " qinghuayuan " exists the pause in 1 second afterwards.It is 5 milliseconds that inquiry " user-average input time " table obtains the average dead time that described user inputs a character.Calculate and be spaced apart 200 normalized input time, exceeded interval threshold, therefore " n " place in the end will carry out real-time cutting along with the input of user; User is follow-up when proceeding to input, and " mengxiang " average dead time of input is still 5 milliseconds, and after input complete " mengxiang ", create again the pause of 0.5 second, have also exceeded interval threshold, carry out cutting equally in real time, because the cutting after " qinghuayuan " has been carried out complete, so secondary cutting does not impact it, then suppose to make marks with space, list entries after cutting is " qinghuayuanmengxiang ", i.e. cutting sequence is respectively " qinghuayuan " and " mengxiang ".
The candidate Chinese character string finding " qinghuayuan " corresponding for cutting sequence " qinghuayuan " and " mengxiang " basis " pinyin-Hanzi " index comprises " Site of Qing Hua Yuan ", " Tsing-Hua University garden " and " fine garden ", and the candidate Chinese character string that " mengxiang " is corresponding comprises " dream " and " dreamland ".The candidate Chinese character string of above-mentioned different cutting sequence is carried out combining as shown in Figure 5, when then current input character string inputs to " qinghuayuanmengxiang ", just comprise candidate Chinese character sequence: " Site of Qing Hua Yuan dream ", " Tsing-Hua University's garden dream ", " fine garden dream ", " Site of Qing Hua Yuan dreamland ", " Tsing-Hua University's garden dreamland " and " fine garden dreamland ", and the Chinese character sequence such as " Tsing-Hua University interprets a dream and thinks ", " blue and white unit dream " can not be obtained.
Then the language model that off-line training is good is called, determine that its respective probable value sorts by the language model scores calculating these candidate Chinese character sequences, determine that the probability of candidate Chinese character sequence " Site of Qing Hua Yuan dream " is the highest, so it is exactly meet the Chinese character sequence that user inputs intention most.
Then using " Site of Qing Hua Yuan dream " as keyword, the search word inquired also sorts according to eigenwert by query search glossarial index, obtain that candidate sequence comprises " Site of Qing Hua Yuan dream ", " Site of Qing Hua Yuan dream with glorious ", " Site of Qing Hua Yuan dream of in palace ", " Site of Qing Hua Yuan Dreams May Come " and " turning one's dream into reality in Site of Qing Hua Yuan " form collection of candidate sequences, then directly can obtain the Chinese character sequence of " Site of Qing Hua Yuan is dreamed of and glory ".
Then search for each candidate sequence in a search engine, after the process operation of query webpage index, summary data and sequence, each candidate sequence all obtains the highest front 3 Search Results of sequence.The current Search Results now also only can searching for the candidate sequence " Site of Qing Hua Yuan dream " meeting user view most, follow-up according to demand real-time search feeding back again.
Collection of candidate sequences and corresponding Search Results are fed back to input method client and show, as shown in Figure 6.Wherein part A is shown input of character string and is met the candidate sequence of user view most, and part B shows each candidate sequence in collection of candidate sequences, the Search Results of the first candidate sequence of C portion shows.
Follow-up, if according to showing, user found that candidate sequence " Site of Qing Hua Yuan dream is with glorious " more meets oneself search intention, now without the need to continuing input, only need keypad upper and lower key selection or mouse-over to candidate sequence " Site of Qing Hua Yuan dream is with glorious ", input method client just can Search Results corresponding to show candidate sequence " Site of Qing Hua Yuan dream with glory ", as shown in Figure 7.
After above-mentioned input prediction and search recommendation process, as long as user normally inputs according to oneself custom, the Search Results of candidate sequence according to this input prediction and candidate sequence can be obtained.Without the need to being carried out the selection of each candidate item by input method, do not click and carry out the operation of upper screen yet, Deng not needing the instruction of further click " search " button triggering searches, carry out searching for and showing result to jump to result of page searching, just directly can obtain the Search Results needed, considerably reduce user operation flow process, improve search efficiency, improve operating experience.
It should be noted that, for embodiment of the method, in order to simple description, therefore it is all expressed as a series of combination of actions, but those skilled in the art should know, the embodiment of the present invention is not by the restriction of described sequence of movement, because according to the embodiment of the present invention, some step can adopt other orders or carry out simultaneously.Secondly, those skilled in the art also should know, the embodiment described in instructions all belongs to preferred embodiment, and involved action might not be that the embodiment of the present invention is necessary.
Embodiment four
On the basis of above-described embodiment, the present embodiment additionally provides a kind of input method system.
With reference to Fig. 8, show the structured flowchart of a kind of input method system embodiment of the present invention, specifically can comprise as lower module:
Receiver module 802, for receiving the current input information of user's input, wherein, described current input information comprises current input character string and time interval information; Modular converter 804, for carrying out cutting to described current input character string in real time according to described time interval information and being converted to collection of candidate sequences corresponding to described current input character string; Search module 806, for searching for respectively each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence; Feedback module 808, for feeding back according to described collection of candidate sequences and corresponding Search Results.
Wherein, this input method system can comprise: server 1 and input method client 2.Then above-mentioned receiver module 802, modular converter 804, search module 806 and feedback module 808 all can be arranged in server 1 (as shown in Figure 9), thus by input method client 2, current input information is uploaded to the receiver module 802 of server 1, and collection of candidate sequences and corresponding Search Results are fed back to input method client 2 by feedback module 808.
In addition, receiver module 802 and modular converter 804 also can be arranged in input method client 2, and search module 806 and feedback module 808 are arranged in server 1 (as shown in Figure 10), be uploaded to server 1 after the modular converter 804 of input method client 2 is converted to collection of candidate sequences, then by the feedback module 808 of server 1, collection of candidate sequences and corresponding Search Results fed back to input method client 2.
In sum, receive the current input information comprising current input character string and time interval information of user's input, input prediction is carried out by current input information, wherein time interval information characterizes the dead time of current input character string in input process, therefore in the input process of user, carrying out cutting to current input character string in real time according to time interval information can be accurate, cutting word occurrence efficiently or character string corresponding to word, improve Input matching degree, thus be converted to collection of candidate sequences accurately, adopt each candidate sequence in this collection of candidate sequences to carry out search respectively again and obtain Search Results, and feed back according to described collection of candidate sequences and corresponding Search Results, directly can obtain by input method the Search Results wanted without the need to inputted search in the search box of search engine, easy and simple to handle, fast, improve search efficiency.
Optionally, described time interval information comprises the dead time of every two adjacent characters in described current input character string; Described modular converter, comprise: cutting submodule, for when detecting that the dead time produced in real time in input process meets pre-conditioned, the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding; Transform subblock, for changing successively each cutting sequence and combine, obtains the collection of candidate sequences that described current input character string is corresponding.
Described current input information also comprises user account information; Described modular converter, also comprises: normalization submodule, for searching the average dead time of respective user according to described user account information; Adopt the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtain the normalization dwell interval produced in real time.
Described cutting submodule, the normalization dwell interval also for producing in real time in described current input character string, compares with pause threshold value successively; When the normalization dwell interval produced in real time in input process is greater than described pause threshold value, determine that the dead time of described real-time generation meets pre-conditioned.
Described server also comprises: pretreatment module, for determining the average dead time of user, specifically for determining user according to described user account information, and collects the history input information of described user; The time of pausing after inputting single character according to user described in described history input Information Statistics, as the average dead time of described user.
Described transform subblock, comprising: search unit, for searching at least one candidate Chinese character string of each cutting sequence from preset index respectively; Assembled unit, for being combined by the candidate Chinese character string of each cutting sequence, obtains each Chinese character sequence that described input of character string is corresponding respectively; Choosing unit, for obtaining the probability of each Chinese character sequence in language model, and choosing described each Chinese character Sequence composition collection of candidate sequences respectively according to described probability.
Optionally, described in choose unit, for sorting according to described probability is descending to each Chinese character sequence, and each Chinese character sequence coming top N is formed collection of candidate sequences as candidate sequence, wherein N is positive integer.
Optionally, described in choose unit, for sorting according to described probability is descending to each Chinese character sequence, choose the Chinese character sequence of maximum probability as keyword; From the index database of search engine, inquiry and described keyword have the search word of correlativity, and sort according to characteristic of correspondence value to the search word inquired; Choose the search word coming front M position and form described collection of candidate sequences respectively as candidate sequence, wherein M is positive integer.
Optionally, described search module, comprising: search submodule, for searching for successively each candidate sequence in described collection of candidate sequences in a search engine; Screening submodule, for the Search Results for each candidate sequence, screen the Search Results coming front X position of described search engine feedback, wherein X is positive integer.
Described system also comprises: display module, for showing respectively described collection of candidate sequences and corresponding Search Results; Described display module comprises: first shows submodule, for showing successively the candidate sequence in described collection of candidate sequences; Second shows submodule, is illustrated in each Search Results corresponding to the first candidate sequence for obtaining and shows.
Described system also comprises: indicating module, choosing instruction, determining the candidate sequence chosen for receiving to described candidate sequence; Then described second shows submodule, and also for each Search Results that the candidate sequence chosen described in obtaining is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is shown.
To sum up, the embodiment of the present invention is distinguished different users by user account information, to dope the candidate sequence meeting this user view most, obtains Search Results accurately and recommends.
For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part illustrates see the part of embodiment of the method.
Each embodiment in this instructions all adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar part mutually see.
Those skilled in the art should understand, the embodiment of the embodiment of the present invention can be provided as method, device or computer program.Therefore, the embodiment of the present invention can adopt the form of complete hardware embodiment, completely software implementation or the embodiment in conjunction with software and hardware aspect.And the embodiment of the present invention can adopt in one or more form wherein including the upper computer program implemented of computer-usable storage medium (including but not limited to magnetic disk memory, CD-ROM, optical memory etc.) of computer usable program code.
The embodiment of the present invention describes with reference to according to the process flow diagram of the method for the embodiment of the present invention, terminal device (system) and computer program and/or block scheme.Should understand can by the combination of the flow process in each flow process in computer program instructions realization flow figure and/or block scheme and/or square frame and process flow diagram and/or block scheme and/or square frame.These computer program instructions can being provided to the processor of multi-purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminal equipment to produce a machine, making the instruction performed by the processor of computing machine or other programmable data processing terminal equipment produce device for realizing the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
These computer program instructions also can be stored in can in the computer-readable memory that works in a specific way of vectoring computer or other programmable data processing terminal equipment, the instruction making to be stored in this computer-readable memory produces the manufacture comprising command device, and this command device realizes the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
These computer program instructions also can be loaded on computing machine or other programmable data processing terminal equipment, make to perform sequence of operations step to produce computer implemented process on computing machine or other programmable terminal equipment, thus the instruction performed on computing machine or other programmable terminal equipment is provided for the step realizing the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
Although described the preferred embodiment of the embodiment of the present invention, those skilled in the art once obtain the basic creative concept of cicada, then can make other change and amendment to these embodiments.So claims are intended to be interpreted as comprising preferred embodiment and falling into all changes and the amendment of embodiment of the present invention scope.
Finally, also it should be noted that, in this article, the such as relational terms of first and second grades and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or terminal device and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or terminal device.When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the terminal device comprising described key element and also there is other identical element.
Above to a kind of searching method based on input prediction provided by the present invention and a kind of input method system, be described in detail, apply specific case herein to set forth principle of the present invention and embodiment, the explanation of above embodiment just understands method of the present invention and core concept thereof 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.
Claims (18)
1. based on a searching method for input prediction, it is characterized in that, comprising:
Receive the current input information of user's input, wherein, described current input information comprises current input character string and time interval information;
In real time cutting is carried out to described current input character string according to described time interval information, and be converted to collection of candidate sequences corresponding to described current input character string;
Each candidate sequence in described collection of candidate sequences is searched for respectively, obtains each self-corresponding Search Results of each candidate sequence, and feed back according to described collection of candidate sequences and corresponding Search Results.
2. method according to claim 1, is characterized in that, described time interval information comprises the dead time of every two adjacent characters in described current input character string;
Describedly in real time according to described time interval information described current input character string carried out to cutting and be converted to collection of candidate sequences, comprising:
When detecting that the dead time produced in real time in input process meets pre-conditioned, the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding;
Each cutting sequence is changed successively and combined, obtains the collection of candidate sequences that described current input character string is corresponding.
3. method according to claim 2, is characterized in that, described current input information also comprises user account information;
Describedly in real time cutting carried out to described current input character string according to described time interval information and before being converted to candidate sequence, also comprise:
The average dead time of respective user is searched according to described user account information;
Adopt the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtain the normalization dwell interval produced in real time.
4. method according to claim 3, is characterized in that, also comprises:
By the normalization dwell interval produced in real time in described current input character string, compare with pause threshold value successively;
When the normalization dwell interval produced in real time in input process is greater than described pause threshold value, determine that the dead time of described real-time generation meets pre-conditioned.
5. method according to claim 3, is characterized in that, also comprises the determining step of the average dead time of user:
Determine user according to described user account information, and collect the history input information of described user;
The time of pausing after inputting single character according to user described in described history input Information Statistics, as the average dead time of described user.
6. method according to claim 2, is characterized in that, changes successively each cutting sequence and combines, and obtains the collection of candidate sequences that described current input character string is corresponding, comprising:
At least one candidate Chinese character string of each cutting sequence is searched respectively from preset index;
The candidate Chinese character string of each cutting sequence is combined, obtains each Chinese character sequence that described input of character string is corresponding respectively;
In language model, obtain the probability of each candidate Chinese character sequence respectively, and choose described each Chinese character Sequence composition collection of candidate sequences respectively according to described probability.
7. method according to claim 6, is characterized in that, describedly chooses described each Chinese character Sequence composition collection of candidate sequences respectively according to described probability, comprising:
Sort according to described probability is descending to each Chinese character sequence, and each Chinese character sequence coming top N is formed collection of candidate sequences as candidate sequence, wherein N is positive integer.
8. method according to claim 6, is characterized in that, describedly chooses described each Chinese character Sequence composition collection of candidate sequences according to described probability, comprising:
Each Chinese character sequence is sorted according to described probability is descending, chooses the Chinese character sequence of maximum probability as keyword;
From the index database of search engine, inquiry and described keyword have the search word of correlativity, and sort according to characteristic of correspondence value to the search word inquired;
Choose the search word coming front M position and form described collection of candidate sequences respectively as candidate sequence, wherein M is positive integer.
9. method according to claim 1, is characterized in that, describedly searches for respectively each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence, comprising:
In a search engine each candidate sequence in described collection of candidate sequences is searched for successively;
For the Search Results of each candidate sequence, screen the Search Results coming front X position of described search engine feedback, wherein X is positive integer.
10. the method according to claim 1 or 9, is characterized in that, also comprises: show respectively described collection of candidate sequences and corresponding Search Results, be specially:
Candidate sequence in described collection of candidate sequences is shown successively;
Obtain and be illustrated in each Search Results corresponding to the first candidate sequence and show.
11. methods according to claim 10, is characterized in that, also comprise:
Receive and instruction is chosen to described candidate sequence, determine the candidate sequence chosen;
Each Search Results that the candidate sequence chosen described in acquisition is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is shown.
12. 1 kinds of input method systems, is characterized in that, comprising:
Receiver module, for receiving the current input information of user's input, wherein, described current input information comprises current input character string and time interval information;
Modular converter, for carrying out cutting to described current input character string in real time according to described time interval information and being converted to collection of candidate sequences corresponding to described current input character string;
Search module, for searching for respectively each candidate sequence in described collection of candidate sequences, obtains each self-corresponding Search Results of each candidate sequence;
Feedback module, for feeding back according to described collection of candidate sequences and corresponding Search Results.
13. systems according to claim 12, is characterized in that, described time interval information comprises the dead time of every two adjacent characters in described current input character string;
Described modular converter, comprising:
Cutting submodule, for when detecting that the dead time produced in real time in input process meets pre-conditioned, the input position place corresponding at described current input character string carries out cutting, obtains each cutting sequence that described current input character string is corresponding;
Transform subblock, for changing successively each cutting sequence and combine, obtains the collection of candidate sequences that described current input character string is corresponding.
14. systems according to claim 13, is characterized in that, described current input information also comprises user account information;
Described modular converter, also comprises:
Normalization submodule, for searching the average dead time of respective user according to described user account information; Adopt the average dead time of described user to be normalized the dead time produced in real time in input process respectively, obtain the normalization dwell interval produced in real time.
15. systems according to claim 14, is characterized in that:
Described cutting submodule, the normalization dwell interval also for producing in real time in described current input character string, compares with pause threshold value successively; When the normalization dwell interval produced in real time in input process is greater than described pause threshold value, determine that the dead time of described real-time generation meets pre-conditioned.
16. systems according to claim 12, is characterized in that, described search module, comprising:
Search submodule, for searching for successively each candidate sequence in described collection of candidate sequences in a search engine;
Screening submodule, for the Search Results for each candidate sequence, screen the Search Results coming front X position of described search engine feedback, wherein X is positive integer.
17. systems according to claim 12 or 16, it is characterized in that, described system also comprises:
Display module, for showing respectively described collection of candidate sequences and corresponding Search Results;
Wherein, described display module, comprising:
First shows submodule, for showing successively the candidate sequence in described collection of candidate sequences;
Second shows submodule, is illustrated in each Search Results corresponding to the first candidate sequence for obtaining and shows.
18. systems according to claim 17, is characterized in that, described system also comprises:
Indicating module, choosing instruction for receiving to described candidate sequence, determining the candidate sequence chosen;
Then described second shows submodule, and also for each Search Results that the candidate sequence chosen described in obtaining is corresponding, each Search Results being illustrated in the first candidate sequence described in replacement corresponding is shown.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410379907.3A CN105335415A (en) | 2014-08-04 | 2014-08-04 | Search method based on input prediction, and input method system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410379907.3A CN105335415A (en) | 2014-08-04 | 2014-08-04 | Search method based on input prediction, and input method system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105335415A true CN105335415A (en) | 2016-02-17 |
Family
ID=55285947
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410379907.3A Pending CN105335415A (en) | 2014-08-04 | 2014-08-04 | Search method based on input prediction, and input method system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105335415A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105786492A (en) * | 2016-02-23 | 2016-07-20 | 浪潮软件集团有限公司 | Method for realizing code prediction prompt by using big data method |
| CN106024013A (en) * | 2016-04-29 | 2016-10-12 | 努比亚技术有限公司 | Voice data searching method and system |
| CN106484135A (en) * | 2016-09-23 | 2017-03-08 | 百度在线网络技术（北京）有限公司 | A kind of method and apparatus for providing input candidate item |
| CN106649640A (en) * | 2016-12-08 | 2017-05-10 | 武汉斗鱼网络科技有限公司 | Instant search method and device |
| CN107219932A (en) * | 2016-03-21 | 2017-09-29 | 阿里巴巴集团控股有限公司 | A kind of input method, device and intelligent terminal |
| CN107305493A (en) * | 2016-04-20 | 2017-10-31 | 谷歌公司 | Graphic keyboard application with integration search |
| CN107678560A (en) * | 2017-08-31 | 2018-02-09 | 科大讯飞股份有限公司 | The candidate result generation method and device of input method, storage medium, electronic equipment |
| CN107766238A (en) * | 2017-09-27 | 2018-03-06 | 北京金山安全软件有限公司 | Software testing method, device, equipment and computer readable storage medium |
| CN108241440A (en) * | 2016-12-27 | 2018-07-03 | 北京搜狗科技发展有限公司 | A kind of candidate word methods of exhibiting and device |
| CN108279783A (en) * | 2017-12-14 | 2018-07-13 | 北京百度网讯科技有限公司 | Method, apparatus, equipment and the computer storage media shielded in a kind of candidate item |
| CN108304367A (en) * | 2017-04-07 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Segmenting method and device |
| CN109901725A (en) * | 2017-12-07 | 2019-06-18 | 北京搜狗科技发展有限公司 | A kind of pinyin string cutting method and device |
| CN110728113A (en) * | 2018-06-28 | 2020-01-24 | 北京金山办公软件股份有限公司 | Information screening method and device of electronic forms and terminal equipment |
| CN112580747A (en) * | 2020-12-29 | 2021-03-30 | 珠海金山网络游戏科技有限公司 | Matching method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101382844A (en) * | 2008-10-24 | 2009-03-11 | 上海埃帕信息科技有限公司 | Method for inputting spacing participle |
| CN102073699A (en) * | 2010-12-20 | 2011-05-25 | 百度在线网络技术（北京）有限公司 | Method, device and equipment for improving search result based on user behaviors |
| CN102495678A (en) * | 2011-11-30 | 2012-06-13 | 左盼 | Information display method and system based on input method |
| CN102799663A (en) * | 2012-07-13 | 2012-11-28 | 深圳市同洲电子股份有限公司 | Input method and input method system |
| CN103631385A (en) * | 2012-08-23 | 2014-03-12 | 北京搜狗科技发展有限公司 | Method and device for screening candidate items in character input |
| CN103869999A (en) * | 2012-12-11 | 2014-06-18 | 百度国际科技（深圳）有限公司 | Method and device for sorting candidate items generated by input method |
-
2014
- 2014-08-04 CN CN201410379907.3A patent/CN105335415A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101382844A (en) * | 2008-10-24 | 2009-03-11 | 上海埃帕信息科技有限公司 | Method for inputting spacing participle |
| CN102073699A (en) * | 2010-12-20 | 2011-05-25 | 百度在线网络技术（北京）有限公司 | Method, device and equipment for improving search result based on user behaviors |
| CN102495678A (en) * | 2011-11-30 | 2012-06-13 | 左盼 | Information display method and system based on input method |
| CN102799663A (en) * | 2012-07-13 | 2012-11-28 | 深圳市同洲电子股份有限公司 | Input method and input method system |
| CN103631385A (en) * | 2012-08-23 | 2014-03-12 | 北京搜狗科技发展有限公司 | Method and device for screening candidate items in character input |
| CN103869999A (en) * | 2012-12-11 | 2014-06-18 | 百度国际科技（深圳）有限公司 | Method and device for sorting candidate items generated by input method |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105786492A (en) * | 2016-02-23 | 2016-07-20 | 浪潮软件集团有限公司 | Method for realizing code prediction prompt by using big data method |
| CN107219932A (en) * | 2016-03-21 | 2017-09-29 | 阿里巴巴集团控股有限公司 | A kind of input method, device and intelligent terminal |
| CN107305493A (en) * | 2016-04-20 | 2017-10-31 | 谷歌公司 | Graphic keyboard application with integration search |
| CN106024013A (en) * | 2016-04-29 | 2016-10-12 | 努比亚技术有限公司 | Voice data searching method and system |
| CN106484135B (en) * | 2016-09-23 | 2019-03-19 | 百度在线网络技术（北京）有限公司 | It is a kind of for provide input candidate item method and apparatus |
| CN106484135A (en) * | 2016-09-23 | 2017-03-08 | 百度在线网络技术（北京）有限公司 | A kind of method and apparatus for providing input candidate item |
| CN106649640B (en) * | 2016-12-08 | 2020-05-12 | 武汉斗鱼网络科技有限公司 | Instant search method and device |
| CN106649640A (en) * | 2016-12-08 | 2017-05-10 | 武汉斗鱼网络科技有限公司 | Instant search method and device |
| CN108241440A (en) * | 2016-12-27 | 2018-07-03 | 北京搜狗科技发展有限公司 | A kind of candidate word methods of exhibiting and device |
| CN108241440B (en) * | 2016-12-27 | 2023-02-17 | 北京搜狗科技发展有限公司 | Candidate word display method and device |
| CN108304367A (en) * | 2017-04-07 | 2018-07-20 | 腾讯科技（深圳）有限公司 | Segmenting method and device |
| WO2018184510A1 (en) * | 2017-04-07 | 2018-10-11 | 腾讯科技（深圳）有限公司 | Word partitioning method and device and storage medium |
| CN108304367B (en) * | 2017-04-07 | 2021-11-26 | 腾讯科技（深圳）有限公司 | Word segmentation method and device |
| CN107678560A (en) * | 2017-08-31 | 2018-02-09 | 科大讯飞股份有限公司 | The candidate result generation method and device of input method, storage medium, electronic equipment |
| CN107766238B (en) * | 2017-09-27 | 2021-04-09 | 北京金山安全软件有限公司 | Software testing method, device, equipment and computer readable storage medium |
| CN107766238A (en) * | 2017-09-27 | 2018-03-06 | 北京金山安全软件有限公司 | Software testing method, device, equipment and computer readable storage medium |
| CN109901725A (en) * | 2017-12-07 | 2019-06-18 | 北京搜狗科技发展有限公司 | A kind of pinyin string cutting method and device |
| CN109901725B (en) * | 2017-12-07 | 2022-05-06 | 北京搜狗科技发展有限公司 | Pinyin string segmentation method and device |
| CN108279783A (en) * | 2017-12-14 | 2018-07-13 | 北京百度网讯科技有限公司 | Method, apparatus, equipment and the computer storage media shielded in a kind of candidate item |
| CN110728113A (en) * | 2018-06-28 | 2020-01-24 | 北京金山办公软件股份有限公司 | Information screening method and device of electronic forms and terminal equipment |
| CN112580747A (en) * | 2020-12-29 | 2021-03-30 | 珠海金山网络游戏科技有限公司 | Matching method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105335415A (en) | Search method based on input prediction, and input method system | |
| CN104933100B (en) | keyword recommendation method and device | |
| CN108763321B (en) | Related entity recommendation method based on large-scale related entity network | |
| CN105045875B (en) | Personalized search and device | |
| CN102890711B (en) | A kind of retrieval ordering method and system | |
| CN102968465B (en) | Network information service platform and the search service method based on this platform thereof | |
| Sugiyama et al. | Refinement of TF-IDF schemes for web pages using their hyperlinked neighboring pages | |
| CN103838833A (en) | Full-text retrieval system based on semantic analysis of relevant words | |
| CN105488024A (en) | Webpage topic sentence extraction method and apparatus | |
| CN102902826B (en) | A kind of image method for quickly retrieving based on reference picture index | |
| CN103593410A (en) | System for search recommendation by means of replacing conceptual terms | |
| CN104965905A (en) | Web page classifying method and apparatus | |
| CN102314452B (en) | A kind of method and system of being undertaken navigating by input method platform | |
| CN103870505A (en) | Query term recommending method and query term recommending system | |
| CN103365839A (en) | Recommendation search method and device for search engines | |
| CN102096717A (en) | Search method and search engine | |
| CN103678576A (en) | Full-text retrieval system based on dynamic semantic analysis | |
| CN101169780A (en) | Semantic ontology retrieval system and method | |
| CN103294815A (en) | Search engine device with various presentation modes based on classification of key words and searching method | |
| CN105808590A (en) | Search engine realization method as well as search method and apparatus | |
| CN103902597A (en) | Method and device for determining search relevant categories corresponding to target keywords | |
| CN105426529A (en) | Image retrieval method and system based on user search intention positioning | |
| CN102081668A (en) | Information retrieval optimizing method based on domain ontology | |
| CN104021125A (en) | Search engine sorting method and system and search engine | |
| CN103136213A (en) | Method and device for providing related words |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160217 |
|
| RJ01 | Rejection of invention patent application after publication |