CN105447080A - Query completion method in community ask-answer search - Google Patents
Query completion method in community ask-answer search Download PDFInfo
- Publication number
- CN105447080A CN105447080A CN201510745059.8A CN201510745059A CN105447080A CN 105447080 A CN105447080 A CN 105447080A CN 201510745059 A CN201510745059 A CN 201510745059A CN 105447080 A CN105447080 A CN 105447080A
- Authority
- CN
- China
- Prior art keywords
- answer
- question
- search
- candidate
- word
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/242—Query formulation
- G06F16/2425—Iterative querying; Query formulation based on the results of a preceding query
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a query completion method in a community ask-answer search, and belongs to the technical field of information retrieval. The method comprises the following steps: step 1: building a cQA database; step 2: building an entity dictionary and a phrase dictionary; step 3: for a query statement input by a user, determining whether the last word is complete, and if not, performing completion on the last word; step 4: for the query statement, sorting questions in the database by using a sorting function, to obtain a list of initial candidate questions; step 5: screening the list according to requirements on divergence, size, local rank preservation and fidelity ; step 6: performing sorting again according to importance and quality of the candidate problems; and step 7: outputting the first N candidate problems after the sorting is performed again, for the user to select. Compared with the prior art, according to the query completion method in a community ask-answer search provided by the present invention, query completion can be achieved in spite of lack of a user search log, thereby solving the problems in the prior art such as similar problem recommendation, an improper size, unchanged relative order and distortion; and the user experience is better than that of an existing search engine.
Description
Technical field
The present invention relates to a kind of inquiry complementing method, particularly relate to a kind of can inquiry complementing method in the search of community's question and answer, auto-complete can be carried out to the user's inquiry in the search of community's question and answer, effectively promote Consumer's Experience, belong to technical field of information retrieval.
Background technology
Along with the development of internet, informationization, the networking process of human society are accelerated greatly, and information retrieval replaces manual information retrieval already and steps into network times.In information retrieval field, inquiry completion technology is very helpful for user search and expressing information demand.User usually inputs a brief query statement instead of a complete problem when retrieving, and for retrieval model, and the information that complete problem provides will be far superior to a brief query statement.Therefore, when user input query statement, how to help user to provide the query statement of complete problem form to have great importance.
At present, inquire about completion technology to make some progress.They are digging user search daily record and web page content information mainly, and the search engine of some main flows additionally provides the recommendation of relevant phrase.For community-based question and answer search (cQA), carry out when lacking user search daily record inquiring about the work that completion is a very challenging property.At present, the inquiry completion technology in the search of community's question and answer is still in the starting stage.
Summary of the invention
The object of the invention is, for the problem of how to carry out inquiring about completion when lacking user search daily record in the search of community's question and answer, to propose a kind of inquiry complementing method based on sequence.Candidate's problem that this method can provide several complete for user is selective, and effective help user carries out inquiry completion, greatly improves Consumer's Experience.
For achieving the above object, the technical solution adopted in the present invention is as follows:
First utilize community's question and answer of crawl to structure cQA database, utilize web page title on wikipedia and common phrase dictionary creation entity and phrase database; Then judge that whether last word of the query statement that user inputs is complete, if end word is imperfect, utilize correlation rule completion end word; Finally, for complete query statement, candidate's problem is initially sorted, screens and sorted.
Concrete technical scheme of the present invention is as follows:
An inquiry complementing method in community's question and answer search, the method comprises the following steps:
Step one, based on question and answer language material build by question and answer to the cQA database formed;
Step 2, build entity dictionary and phrase dictionary based on existing encyclopaedic knowledge resource and dictionary resources;
Step 3, the query statement inputted for user, judge that whether last word of query statement is complete, if end word is imperfect, go to step four; Otherwise, go to step five;
Step 4, the query statement of input is carried out to end word completion and obtains complete query statement;
Step 5, by adopting ranking functions to sort to the problem in database, the initial candidate's problem list selected for user is obtained to query statement;
As preferably, in order to improve retrieval effectiveness, in ranking functions, have employed the model of the level and smooth method of linear interpolation and statistical translation, specific as follows:
p(w|(q，a))＝(1-λ)p
mx(w|(q，a))+λp
ml(w|C)(2)
Wherein, q is the query statement of user, and w is each word in query statement, and C={ (q, a)
1, (q, a)
2..., (q, a)
lrepresent question and answer pair in cQA database, (q, a)
ii-th question and answer pair, q problem of representation, a represents answer; λ is smoothing parameter, p
ml(w|C) given C is represented, the conditional probability of word w; p
ml(w|q) given problem q is represented, the conditional probability of word w; P (w|t) represents the word t in given problem, the conditional probability of word w; p
ml(t|q) given problem q is represented, the conditional probability of word t; p
ml(w|a) given answer a is represented, the conditional probability of word w; And meet alpha+beta+γ=1.
Step 6, the initial candidate problem list obtained step 4 screen according to the requirement of diversity, size appropriateness, local isotonicity and fidelity;
As preferably, the described requirement according to diversity, size appropriateness, local isotonicity and fidelity is carried out screening and is realized by the following method:
(1) in order to meet the requirement of diversity, utilize the method for topic model (topicmodel, LDA) to be that each candidate's problem distributes theme, and candidate's problem that restriction returns should come from different themes;
(2) in order to meet the requirement of size appropriateness, limiting the length returning candidate's problem and can not exceed some threshold value s;
(3) in order to the requirement of satisfied local isotonicity, the relative ranks of restriction query terms can change, but the partial order of the entity identified based on entity dictionary and phrase dictionary in query statement and phrase can not change;
(4) in order to meet the requirement of fidelity, the word of user's input must in restriction candidate problem, be comprised;
As preferably, described threshold value s is 62 characters.
As preferably, the entity identified based on entity dictionary and phrase dictionary in described query statement and phrase adopt the method for maximum String matching to identify.
Step 7, the candidate's problem list after screening to be sorted according to the importance of candidate's problem and quality order from high to low again;
As preferably, described importance adopts technorati authority that in cQA database, question and answer are right and user's degree of attentiveness to characterize.
As preferably, the value of described importance is by calculating being added together by equal weight after technorati authority and the regularization of user's degree of attentiveness.
As preferably, described quality adopts the spelling in text, grammar mistake characterizes.
As preferably, the value of described quality is calculated by the counting how many times occurred mistake.
As preferably, further, in order to the difference helping user to identify candidate's problem of output fast, each candidate's the very corn of a subject word is added to bulk processing, does delete processing to invalid title, thus promote Consumer's Experience.
Step 8, the front N bar candidate problem after sorting again that exports are selected for user.
Beneficial effect
Compared with common inquiry complementing method, the present invention can carry out inquiry completion when lacking user search daily record in the search of community's question and answer, adopt the thought recommended candidate problem based on sequence, meet the every demand in cQA search, overcome in traditional directory complementing method and recommend Similar Problems, size is not inconsistent, relative ranks is constant, the problem of distortion.The present invention can help user to carry out inquiry completion effectively, and Consumer's Experience is better than existing search engine result of use, and meanwhile, the present invention can be combined to obtain better effect with existing inquiry complementing method flexibly.
Accompanying drawing explanation
Fig. 1 is the inquiry complementing method schematic flow sheet in the question and answer search of a kind of community of the embodiment of the present invention.
Fig. 2 is inquiry completion exemplary graph (a: weak effect; B: effective).
Fig. 3 is inquiry completion comparative result (a:Yahoo; B:Google; C: the method that the present invention proposes)
Embodiment
Below in conjunction with embodiment and accompanying drawing, the specific embodiment of the present invention is described in further detail.
Inquiry complementing method in a kind of community of the present invention question and answer search, as shown in Figure 1, comprises the following steps:
One, build by question and answer the cQA database formed based on question and answer language material.
Answer crawls language material (problem and corresponding answer), constitute one and be surrounded by 6,345, the cQA database that 786 question and answer are right.Based on this database, we have developed the system of a cQA search, the function of search and inquiry completion is provided.
Two, entity dictionary and phrase dictionary is built based on existing encyclopaedic knowledge resource (as Baidupedia or wikipedia etc.) and dictionary resources.
In order to identify the entity in user's query statement, we identify by the method for a maximum String matching of the Dictionary use be made up of entity.In the present embodiment, because the page title of wikipedia can presentation-entity name, so we adopt wikipedia web page title to build entity dictionary.For the identification of phrase, we make to use the same method, and namely utilize the method for the maximum String matching of Dictionary use; And for the dictionary of phrase, we utilize the dictionary of conventional English phrase.
Three, for the query statement of user's input, judge that whether last word of query statement is complete, if end word is imperfect, go to step four; Otherwise, go to step five.
Four, end word completion is carried out to the query statement of input and obtain complete query statement.
The method of completion end word, performing step is as follows:
In default of search daily record, the famous correlation rule learning algorithm (as Apriori algorithm) of our usage data excavation applications goes to school acquistion to correlation rule at language material, the correlation rule that these study obtain and the frequency of occurrences thereof are similar to the search daily record in search technique, thus the word completion technology in search technique just can be utilized to realize.
Five, by adopting ranking functions to sort to the problem in database, the initial candidate's problem list selected for user is obtained to query statement.
In general search, in search daily record, the frequency of occurrences of query statement is a key factor of inquiry completion, but owing to lacking search daily record, we adopt the method for ranked candidate problem.Here, we consider that the problem owing to much having different semantic structure and the form of expression all have expressed equivalent, such as, problem " you can tell me the difference of iphone5 and iphone5s? " " what the difference of iphone5 and iphone5s is? " the two has different syntactic structures and but expresses the identical meaning, therefore can not use the frequency of occurrences of problem to sort.Consider these, in order to obtain an initial sorted lists, we use a ranking functions to sort to the problem in database.Formula is as follows:
p(w|(q，a))＝(1-λ)p
mx(w|(q，a))+λp
ml(w|C)(2)
Wherein, q is the query statement of user, and w is that each word in query statement (for Chinese, needs first to do Chinese word segmentation; For English, adopt space-separated words), and C={ (q, a)
1, (q, a)
2..., (q, a)
lrepresent question and answer pair in cQA database, (q, a)
ii-th question and answer pair, q problem of representation, a represents answer; λ is smoothing parameter, p
ml(w|C) given C is represented, the conditional probability of word w; p
ml(w|q) given problem q is represented, the conditional probability of word w; P (w|t) represents the word t in given problem, the conditional probability of word w; p
ml(t|q) given problem q is represented, the conditional probability of word t; p
ml(w|a) given answer a is represented, the conditional probability of word w; And meet alpha+beta+γ=1.
By we can calculate problem in cQA database to the possibility of generated query statement with superior function, whereby problem is sorted, generate initial candidate problem list.
What be necessary explanation is, in the present embodiment, we portray the semantic relation between word and word by Using statistics translation model, employ linear interpolation smoothing method simultaneously and the advantage of background language model and translation model are combined, thus are conducive to improving sequence effect.
Six, the initial candidate problem list that step 4 obtains is screened according to the requirement of diversity, size appropriateness, local isotonicity and fidelity.
Four demands below inquiry completion technology demand fulfillment in cQA search:
(1) diversity: as shown in Fig. 2 (a), user's input " iphone3gsiphone3g " clicks " search " button afterwards, the recommendation problem as shown in drop-down list that system provides mostly have expressed the similar meaning, this can waste the time that user browses, reduce Consumer's Experience effect, and Fig. 2 (b) has good diversity;
(2) size appropriateness: as shown in Fig. 2 (a), last problem of recommending, beyond the size of Suggestion box, so not only affects attractive in appearance, reduces Consumer's Experience effect simultaneously, and the problem size appropriateness of recommending in Fig. 2 (b);
(3) local isotonicity: for entity and phrase, keep their partial order to be necessary, and there is no need the relative ranks that keeps in candidate's problem between query terms;
(4) fidelity: inquiry completion technology can not lose the primitive meaning of the query statement of user's input.
For needing above, we screen initial sorted lists:
(1) in order to meet the requirement of diversity, we utilize the method for topic model (topicmodel, LDA) to be that each candidate's problem distributes theme, and candidate's problem that restriction returns should come from different themes;
(2) in order to meet the requirement of size appropriateness, we limit the length returning candidate's problem can not exceed some threshold values, and here, our threshold value of setting is 62 characters;
(3) in order to the requirement of satisfied local isotonicity, the relative ranks that we limit query terms can change, but the partial order of the entity identified based on entity dictionary and phrase dictionary in query statement and phrase can not change;
(4) in order to meet the requirement of fidelity, we limit in candidate's problem the word that must comprise user's input.
Seven, the candidate's problem list after screening is sorted according to the importance of candidate's problem and quality order from high to low again.
After initially sorting and screening, we can only obtain a coarse problem sorted lists.In sequencer procedure, we assess the importance of candidate's problem and quality again.
Importance: the technorati authority that we use question and answer right and user's degree of attentiveness are to weigh the importance of problem.The quantity of the best problem that (1) user answers can be used for weighing the technorati authority of this user, and therefore, we use the technorati authority summation of asker and answerer as the right technorati authority of these question and answer.(2) question and answer can represent the degree of attentiveness of user to this problem to comprised whole answer quantity.To sum up, we will be added together the importance of problem of representation by equal weight after these two regularizations.
Quality: cQA data are issued by user, and text ubiquity some spelling, grammar mistakes etc. that user issues.Therefore, the number of times that our mistake in occurs weighs the quality of problem.
After considering the importance of problem and quality, we sort to the list after screening again, simultaneously, in order to the difference helping user to identify different candidate's problem fast, we have carried out adding bulk processing to each candidate's the very corn of a subject word (such as interrogative, verb and noun), improve Consumer's Experience like this.As shown in Fig. 3 (c), for the core word in candidate's problem, we have all carried out adding bulk processing, facilitate user to identify like this.In addition, for invalid title, we do delete processing.
Eight, the front N bar candidate problem exported after sequence is again selected for user.
Test findings
Effective in order to verify the inquiry complementing method in cQA search that the present embodiment provides, we to sample 25 problems from data centralization, select the query statement building this problem containing 2,3,4,5 keywords, have 100 query statements like this from these problems.Such as, for problem " whatisthedifferencebetweeniphone5andiphone5s? ", its corresponding query statement is " iphone5 ", " iphone5s ", " iphone55sdifference ", " iphone55swhatdifference ".For each problem, whether comprise this problem in the recommendation list that we use its respective queries statement checking system to return or there is the problem of same meaning, if comprised, mark its sorting position.Recall rate result is as shown in the table:
Table 1 recall rate of the present invention (%)
| Keyword number contained by query statement | Front 1 | Front 2 | Front 6 | Front 10 |
| 2 | 20 | 28 | 44 | 72 |
| 3 | 52 | 68 | 80 | 80 |
| 4 | 72 | 80 | 84 | 84 |
| 5 | 76 | 80 | 84 | 84 |
Experimental result shows, when input 4 keywords, front 2 recommendation problems are the possibilities desired by user is 80%.This demonstrate that the validity of put forward the methods of the present invention.
In fig. 3, the method that we use the present invention to propose compares main flow search engine (Google and Yahoo) with two compares, Fig. 3 (a) and 3 (b) all maintain the relative ranks between query terms, the question sentence of completion is fluently unobstructed not, and user is often more random when inputting, can not emphasize the relative ranks of word, this just needs inquiry completion system query terms can be organized into more clear and more coherent question sentence.The method that Fig. 3 (c) proposes for the present invention, result can find out that method of the present invention is better than them.These two search engines have a large amount of search daily records and Internet resources, and therefore, we it is contemplated that method of the present invention is when having same asset, and the effect of inquiry completion is necessarily better than existing search engine.
More than show and describe ultimate principle of the present invention and principal character and advantage of the present invention.The technician of the industry should understand; the present invention is not restricted to the described embodiments; what describe in above-described embodiment and instructions just illustrates principle of the present invention; without departing from the spirit and scope of the present invention; the present invention also has various changes and modifications; these changes and improvements are all in the claimed scope of the invention, and application claims protection domain is defined by appending claims and equivalent thereof.
Claims (10)
1. the inquiry complementing method in the search of community's question and answer, is characterized in that:
Step one, based on question and answer language material build by question and answer to the cQA database formed;
Step 2, build entity dictionary and phrase dictionary based on existing encyclopaedic knowledge resource and dictionary resources;
Step 3, the query statement inputted for user, judge that whether last word of query statement is complete, if end word is imperfect, go to step four; Otherwise, go to step five;
Step 4, the query statement of input is carried out to end word completion and obtains complete query statement;
Step 5, by adopting ranking functions to sort to the problem in database, the initial candidate's problem list selected for user is obtained to query statement;
Step 6, the initial candidate problem list obtained step 4 screen according to the requirement of diversity, size appropriateness, local isotonicity and fidelity;
Step 7, the candidate's problem list after screening to be sorted according to the importance of candidate's problem and quality order from high to low again;
Step 8, the front N bar candidate problem after sorting again that exports are selected for user.
2. the inquiry complementing method in the search of a kind of community according to claim 1 question and answer, is characterized in that, in order to improve retrieval effectiveness, have employed the model of the level and smooth method of linear interpolation and statistical translation in ranking functions described in step 5, specific as follows:
p(w|(q，a))＝(1-λ)p
mx(w|(q，a))+λp
ml(w|C)(2)
Wherein, q is the query statement of user, and w is each word in query statement, and C={ (q, a)
1, (q, a)
2..., (q, a)
lrepresent question and answer pair in cQA database, (q, a)
ii-th question and answer pair, q problem of representation, a represents answer; λ is smoothing parameter, p
ml(w|C) given C is represented, the conditional probability of word w; p
ml(w|q) given problem q is represented, the conditional probability of word w; P (w|t) represents the word t in given problem, the conditional probability of word w; p
ml(t|q) given problem q is represented, the conditional probability of word t; p
ml(w|a) given answer a is represented, the conditional probability of word w; And meet alpha+beta+γ=1.
3. the inquiry complementing method in the search of a kind of community according to claim 1 question and answer, is characterized in that, the described requirement according to diversity, size appropriateness, local isotonicity and fidelity carries out screening realization by the following method:
(1) in order to meet the requirement of diversity, utilize the method for topic model (topicmodel, LDA) to be that each candidate's problem distributes theme, and candidate's problem that restriction returns should come from different themes;
(2) in order to meet the requirement of size appropriateness, limiting the length returning candidate's problem and can not exceed some threshold value s;
(3) in order to the requirement of satisfied local isotonicity, the relative ranks of restriction query terms can change, but the partial order of the entity identified based on entity dictionary and phrase dictionary in query statement and phrase can not change;
(4) in order to meet the requirement of fidelity, the word of user's input must in restriction candidate problem, be comprised.
4. the inquiry complementing method in the search of a kind of community according to claim 3 question and answer, it is characterized in that, described threshold value s is 62 characters.
5. the inquiry complementing method in the search of a kind of community according to claim 3 question and answer, it is characterized in that, the entity identified based on entity dictionary and phrase dictionary in described query statement and phrase adopt the method for maximum String matching to identify.
6. the inquiry complementing method in the search of a kind of community according to claim 1 question and answer, is characterized in that, importance described in step 7 adopts the technorati authority and user's degree of attentiveness sign that in cQA database, question and answer are right.
7. the inquiry complementing method in the search of a kind of community according to claim 6 question and answer, it is characterized in that, the value of described importance is by calculating being added together by equal weight after technorati authority and the regularization of user's degree of attentiveness.
8. the inquiry complementing method in the search of a kind of community according to claim 6 question and answer, is characterized in that, the spelling in described quality employing text, grammar mistake characterize.
9. the inquiry complementing method in the search of a kind of community according to claim 8 question and answer, it is characterized in that, the value of described quality is calculated by the counting how many times occurred mistake.
10. the inquiry complementing method in the search of a kind of community according to claim 1 question and answer, it is characterized in that, in order to the difference helping user to identify candidate's problem of output fast, also need each candidate's the very corn of a subject word is added to bulk processing, does delete processing to invalid title after described step 7 completes, thus promote Consumer's Experience.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510745059.8A CN105447080B (en) | 2015-11-05 | 2015-11-05 | A kind of inquiry complementing method in community's question and answer search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510745059.8A CN105447080B (en) | 2015-11-05 | 2015-11-05 | A kind of inquiry complementing method in community's question and answer search |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105447080A true CN105447080A (en) | 2016-03-30 |
| CN105447080B CN105447080B (en) | 2018-10-26 |
Family
ID=55557258
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510745059.8A Active CN105447080B (en) | 2015-11-05 | 2015-11-05 | A kind of inquiry complementing method in community's question and answer search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105447080B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106294656A (en) * | 2016-08-04 | 2017-01-04 | 武汉大学 | A kind of map locating keyword is to the method for relevant issues |
| CN107016561A (en) * | 2016-10-28 | 2017-08-04 | 阿里巴巴集团控股有限公司 | A kind of information processing method and device |
| CN107092609A (en) * | 2016-05-10 | 2017-08-25 | 口碑控股有限公司 | A kind of information-pushing method and device |
| CN108074116A (en) * | 2016-11-09 | 2018-05-25 | 阿里巴巴集团控股有限公司 | Information providing method and device |
| CN108509476A (en) * | 2017-09-30 | 2018-09-07 | 平安科技（深圳）有限公司 | Problem associates method for pushing, electronic device and computer readable storage medium |
| CN108763251A (en) * | 2018-04-02 | 2018-11-06 | 阿里巴巴集团控股有限公司 | The personalized recommendation method and device and electronic equipment of core body product |
| CN109828981A (en) * | 2017-11-22 | 2019-05-31 | 阿里巴巴集团控股有限公司 | A kind of data processing method and calculate equipment |
| CN110750704A (en) * | 2019-10-23 | 2020-02-04 | 深圳计算科学研究院 | Method and device for automatically completing query |
| CN111209382A (en) * | 2020-01-03 | 2020-05-29 | 联想(北京)有限公司 | Content processing method, device, electronic equipment and medium |
| CN116955577A (en) * | 2023-09-21 | 2023-10-27 | 四川中电启明星信息技术有限公司 | Intelligent question-answering system based on content retrieval |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2181406A1 (en) * | 2007-07-11 | 2010-05-05 | Koninklijke Philips Electronics N.V. | Method of operating an information retrieval system |
| CN102073736A (en) * | 2011-01-20 | 2011-05-25 | 百度在线网络技术（北京）有限公司 | Method and system for searching difficult word |
| CN104123322A (en) * | 2013-04-28 | 2014-10-29 | 百度在线网络技术（北京）有限公司 | Method and device for obtaining related question corresponding to input question based on synonymy processing |
-
2015
- 2015-11-05 CN CN201510745059.8A patent/CN105447080B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2181406A1 (en) * | 2007-07-11 | 2010-05-05 | Koninklijke Philips Electronics N.V. | Method of operating an information retrieval system |
| CN102073736A (en) * | 2011-01-20 | 2011-05-25 | 百度在线网络技术（北京）有限公司 | Method and system for searching difficult word |
| CN104123322A (en) * | 2013-04-28 | 2014-10-29 | 百度在线网络技术（北京）有限公司 | Method and device for obtaining related question corresponding to input question based on synonymy processing |
Non-Patent Citations (1)
| Title |
|---|
| 田宇辰: ""专业搜索引擎的无日志查询推荐机制研究及实现"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092609B (en) * | 2016-05-10 | 2021-04-02 | 口碑控股有限公司 | Information pushing method and device |
| CN107092609A (en) * | 2016-05-10 | 2017-08-25 | 口碑控股有限公司 | A kind of information-pushing method and device |
| CN106294656B (en) * | 2016-08-04 | 2019-03-19 | 武汉大学 | A kind of method of map locating keyword to relevant issues |
| CN106294656A (en) * | 2016-08-04 | 2017-01-04 | 武汉大学 | A kind of map locating keyword is to the method for relevant issues |
| CN107016561A (en) * | 2016-10-28 | 2017-08-04 | 阿里巴巴集团控股有限公司 | A kind of information processing method and device |
| CN107016561B (en) * | 2016-10-28 | 2020-10-20 | 创新先进技术有限公司 | Information processing method and device |
| CN108074116A (en) * | 2016-11-09 | 2018-05-25 | 阿里巴巴集团控股有限公司 | Information providing method and device |
| CN108074116B (en) * | 2016-11-09 | 2022-02-22 | 阿里巴巴集团控股有限公司 | Information providing method and device |
| CN108509476A (en) * | 2017-09-30 | 2018-09-07 | 平安科技（深圳）有限公司 | Problem associates method for pushing, electronic device and computer readable storage medium |
| CN109828981B (en) * | 2017-11-22 | 2023-05-23 | 阿里巴巴集团控股有限公司 | Data processing method and computing device |
| CN109828981A (en) * | 2017-11-22 | 2019-05-31 | 阿里巴巴集团控股有限公司 | A kind of data processing method and calculate equipment |
| CN108763251B (en) * | 2018-04-02 | 2021-06-01 | 创新先进技术有限公司 | Personalized recommendation method and device for nuclear product and electronic equipment |
| CN108763251A (en) * | 2018-04-02 | 2018-11-06 | 阿里巴巴集团控股有限公司 | The personalized recommendation method and device and electronic equipment of core body product |
| CN110750704A (en) * | 2019-10-23 | 2020-02-04 | 深圳计算科学研究院 | Method and device for automatically completing query |
| CN110750704B (en) * | 2019-10-23 | 2022-03-11 | 深圳计算科学研究院 | Method and device for automatically completing query |
| CN111209382A (en) * | 2020-01-03 | 2020-05-29 | 联想(北京)有限公司 | Content processing method, device, electronic equipment and medium |
| CN111209382B (en) * | 2020-01-03 | 2024-02-27 | 联想(北京)有限公司 | Content processing method, device, electronic equipment and medium |
| CN116955577A (en) * | 2023-09-21 | 2023-10-27 | 四川中电启明星信息技术有限公司 | Intelligent question-answering system based on content retrieval |
| CN116955577B (en) * | 2023-09-21 | 2023-12-15 | 四川中电启明星信息技术有限公司 | Intelligent question-answering system based on content retrieval |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105447080B (en) | 2018-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105447080A (en) | Query completion method in community ask-answer search | |
| CN106844658B (en) | Automatic construction method and system of Chinese text knowledge graph | |
| CN104484431B (en) | A kind of multi-source Personalize News webpage recommending method based on domain body | |
| Thakkar et al. | Graph-based algorithms for text summarization | |
| CN107180045B (en) | Method for extracting geographic entity relation contained in internet text | |
| CN104978314B (en) | Media content recommendations method and device | |
| CN108052593A (en) | A kind of subject key words extracting method based on descriptor vector sum network structure | |
| Alexander et al. | Natural language web interface for database (NLWIDB) | |
| Carley et al. | Data-to-model: a mixed initiative approach for rapid ethnographic assessment | |
| CN112925901B (en) | Evaluation resource recommendation method for assisting online questionnaire evaluation and application thereof | |
| Moreno-Ortiz et al. | Design and validation of annotation schemas for aspect-based sentiment analysis in the tourism sector | |
| CN109522396B (en) | Knowledge processing method and system for national defense science and technology field | |
| CN112667821A (en) | Military knowledge graph construction and question-answering method based on encyclopedic data | |
| Carrion et al. | A taxonomy generation tool for semantic visual analysis of large corpus of documents | |
| Staar et al. | Corpus processing service: a knowledge graph platform to perform deep data exploration on corpora | |
| Hasanati et al. | Implementation of support vector machine with lexicon based for sentimenT ANALYSIS ON TWITter | |
| CN113239143A (en) | Power transmission and transformation equipment fault processing method and system fusing power grid fault case base | |
| Kadry et al. | On the improvement of weighted page content rank | |
| Wasim et al. | Extracting and modeling user interests based on social media | |
| Ueyama et al. | Automated construction and evaluation of Japanese Web-based reference corpora | |
| Xu et al. | Objectivity and moral judgment in us news narratives: A natural language processing analysis of ‘culture war’coverage | |
| Kenekayoro et al. | Motivation for hyperlink creation using inter-page relationships | |
| Yilahun et al. | Ontology expansion based on UWN reusability | |
| Sert et al. | Temptracker: a service oriented temporal natural language processing based tool for document data characterization and social network analysis. | |
| CN117909451B (en) | Question and answer result tracing method, device, equipment, medium and program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |