JP5780617B2 - Method and apparatus for acquiring / searching related knowledge information - Google Patents

Description

  The present invention relates to the field of Internet communication technology, and more particularly to a method and apparatus for acquiring / retrieving related knowledge information.

  The present application is a Chinese patent application no. Based on the benefit of priority according to 2011010081274.4 (the filing date is March 31, 2011, and the title of the invention is “method and apparatus for acquiring / searching related knowledge information”), the benefit is obtained.

  With the rapid development of Internet technology, acquiring information and communicating with each other over the Internet is already part of people's daily life. The knowledge question and answer system is a system that uses the communication function to acquire information. Users can submit various questions to the knowledge question and answer system through a web page, look up the status of the submitted question, and Determine which answer to adopt based on the status of the answer. Other users can visit this web page to look up questions and answer based on their preferences and knowledge. However, users can only obtain the necessary knowledge information after submitting a question to the knowledge and answer system, and then waiting for other users to answer this question, so if the user suddenly needs an answer In addition, related knowledge information cannot be provided rapidly and correctly.

  The present invention provides a method and apparatus for acquiring and retrieving related knowledge information so as to provide related knowledge information quickly and correctly.

The specific technical plan is as follows.
Analyzing the search log to find a hot search request query that has a questionable demand Step A, forming a question using the query found in Step A, and publishing it on a page layout of a knowledge question and answer platform, the knowledge A method for acquiring linked knowledge information including a step C of acquiring related knowledge information of the question through a page layout of a question and answer platform.

  The step A specifically identifies a query having a questionable demand in the search log and determines a hot query with the query with a questionable demand, or determines a hot query in the search log, Identifying a query having a questionable demand in the query, or identifying a query having a questionable demand in the search log, and determining a hot query in the search log, and determining a query having the identified questionable demand as a hot Taking a crossed set of queries.

  Further, the step of identifying the query having the questionable demand specifically includes a step of separating the query based on the meaning of the word, and matching each word obtained by the word separation processing in the question attribute database. Determining a question tendency value for each word; obtaining a question tendency value for the query by combining the question tendency values for each word; and whether the question tendency value for the query exceeds a preset question demand threshold Determining that there is a questionable demand for the query; otherwise, determining that there is no questionable demand for the query, the questionable attribute database includes a data discovery scheme or Each word obtained through an artificial arrangement method and a question tendency value corresponding to each word are stored.

  The question tendency value corresponding to the word is determined by factors such as whether or not the word is a question word or the relation between the word and the question word. Specifically, the step of determining the hot query includes a step of performing clustering based on relevance for the query to obtain each query set, and a search frequency for each query for each query set, and a search for each query set. Determining a frequency, determining a query set whose search frequency exceeds a preset hot frequency as a hot query set, selecting one query from the hot query set, and setting it as a hot query including.

  The step of forming a question using the query found in Step A in Step B specifically includes attaching a part-of-speech label to a word that has been subjected to word separation processing based on the meaning of the word for the found query. In accordance with the step and the part-of-speech label attached, the word that has been subjected to the word separation process is compared with the grammar of the question sentence that has been set in advance, and a word that is missing from the word that has undergone the word separation process is added, Combining questions that fit the grammar.

  The step of publishing the question on the page layout of the knowledge question and answer platform is to specifically select one ID from a set of preset simulated question IDs and use this ID to simulate the user to know the question. Including the step of publishing on the page layout of the question and answer platform, the IDs in the set of simulated question IDs are silently regarded as registered user IDs by the knowledge and answer platform. Preferably, in the step C, the related knowledge information obtained by the user answering the question is acquired from the page layout of the knowledge question and answer platform, and a good quality answer is determined from the related knowledge information.

  The method further includes that if the presentation time of the question on the page layout of the knowledge question and answer platform is longer than a preset off time, any relevant knowledge information of the question has not yet appeared, or the quality of the question If an answer has not yet appeared, the method includes turning off a page layout on which the question exists on the knowledge question / answer platform.

  Accepting a query entered by a user, and searching for a page layout that matches the keyword of the query, and when the knowledge and answer platform searches for a page layout that matches the keyword of the query, the knowledge A method of searching related knowledge information based on a method of acquiring the related knowledge information, wherein a page layout that matches the keyword of the query on a question and answer platform is included in a search result of the query and returned to the user.

  Query search means for analyzing a search log to find a hot query having a questionable demand, question formation means for forming a question using the hot query found by the previous term query discovery means, and page layout of the question to the knowledge and answer platform And a knowledge acquisition means for acquiring related knowledge information of the question in a page layout of the knowledge question and answer platform.

  Specifically, the query discovery means includes a demand identification sub-unit for identifying and outputting a query having a questionable demand in the input query, and a hot confirmation sub-unit for determining and outputting a hot query with the input query. The input of the demand identification sub-means is a query in a search log, the input of the hot confirmation sub-means is the output of the demand identification sub-means, and the output of the hot confirmation sub-means is the hot with the questionable demand Or the input of the hot confirmation sub-means is a query of a search log, the input of the demand identification sub-means is the output of the hot confirmation sub-means, and the output of the demand identification sub-means is the question There is a demand hot query, or the input of the demand identification sub means is a query in a search log, and the input of the hot confirmation sub means is also a query in a search log, For, the apparatus further, the hot deterministic sub unit takes intersection set of the demand identification sub unit includes a cross-set processing sub section for outputting a hot queries in doubt demand.

  Specifically, the demand identifying sub-means matches a word separation processing block that performs word separation processing of a query based on the meaning of the word and each word obtained by the word separation processing in a question attribute database, and A block with a word point for determining a question tendency value, a block with a query point for obtaining a question tendency value of the inputted query by combining the question tendency values of the respective words, and a question tendency value of the inputted query are preset. A demand determination block for determining whether or not there is an interrogative demand for the input query, otherwise determining that there is no interrogative demand for the input query. In the question attribute database, each word and each word obtained through a data discovery method or an artificial arrangement method The corresponding question trend values are stored.

  Specifically, the question tendency value corresponding to the word is determined by factors such as whether or not the word is a question word, or a relation between the word and the question word. In addition, the hot determination sub-unit specifically performs clustering based on relevance for the query, and combines the clustering processing block for obtaining each query set and the search frequency of each query for each query set, for each query set. A frequency statistical block for determining the search frequency, a hot set determination block for determining a query set whose search frequency exceeds a preset hot frequency as a hot query set, and one query from each of the hot query sets And a hot query confirmation block as a hot query.

  Specifically, the question forming means includes a part-of-speech identification code sub-means for attaching a part-of-speech label to a word obtained by performing word separation processing based on the meaning of the word for the query found by the query finding means, and a part-of-speech label attached In accordance with the above, the word that has been subjected to the word separation processing is compared with the grammar of the question sentence that has been set in advance, the word that is missing from the word that has undergone the word separation processing is added, and the question that conforms to the grammar of the question sentence is combined And a text combination sub-means.

  Specifically, the question announcement means selects one ID from a set of preset simulated question IDs, uses the selected ID to simulate a user, and asks questions formed by the question formation means as a knowledge question-answer platform The IDs in the set of simulated question IDs are silently registered as registered user IDs by the knowledge question and answer platform.

  Specifically, the knowledge acquisition means acquires related knowledge information for a user to answer the question using the page layout of the knowledge question / answer platform, and determines a good answer from the related knowledge information. Furthermore, if the presentation time of the question on the page layout of the knowledge and answer platform is longer than a preset off time, any related knowledge information of the question has not yet appeared, or a good answer of the question has not yet appeared. If not, page layout maintenance means for turning off the page layout on which the question exists on the knowledge question / answer platform is included.

  A device for acquiring the related knowledge information, a user interface means for accepting a query input by a user, a page layout that matches a keyword of the query, and a device for acquiring the related knowledge information are provided in a knowledge-answer platform. When a page layout that matches the keyword of the query is searched for in the page layout in which the question is announced, the page layout search means is included for causing the user to return the searched page layout to the search result of the query.

  According to the above technical solution, the present invention analyzes a search log to find a hot query that has a questionable demand, forms a question using the found query, and publishes it on the page layout of the knowledge and answer platform. When the user has a related question, the search engine causes the knowledge question and answer platform to return the page layout in which this question exists to the user so that the user obtains relevant knowledge information for the question in this page layout. That is, the present invention can quickly and correctly provide related knowledge information existing in the knowledge and answer platform through the search engine, and the user registers in the knowledge and answer platform, publishes the question, and waits for the question to be answered. It is no longer necessary to acquire relevant knowledge information.

4 is a flowchart of a method for acquiring related knowledge information provided in the first embodiment of the present invention. 4 is a flowchart of a method for determining a hot query provided in a second embodiment of the present invention. It is a flowchart of the method of searching the related knowledge information provided in the 3rd Example of this invention. It is a block diagram of the apparatus which acquires the related knowledge information provided in 4th Example of this invention. (A), (b), (c) are three block diagrams of the query finding means provided in the fourth embodiment of the present invention. It is a block diagram of the demand identification sub means provided in the fourth embodiment of the present invention. It is a block diagram of the hot determination sub means provided in the fourth embodiment of the present invention. It is a block diagram of the apparatus which searches the related knowledge information provided in Example 5 of this invention.

  In order to clarify the objects, technical solutions, and advantages of the present invention, the present invention will be described below in detail with reference to the drawings and specific embodiments.

First Embodiment FIG. 1 is a flowchart of a method for acquiring related knowledge information provided in a first embodiment of the present invention. As shown in FIG. 1, the method includes the following steps. In step 101, the search log is analyzed to find a hot query with questionable demand.

  The search log is periodically analyzed, the search log of the current period is extracted, and a hot query having a questionable demand is found by using the extracted search log in the current period. The period for analyzing the search log may be set flexibly. For example, a hot query having a questionable demand may be found from the search log of the current day as a cycle.

  This step is actually divided into two operations. One operation is to identify whether the query in the search log is a questionable demand, and another operation is to determine the hot query. These two operations may be performed in any order, and may be performed synchronously in parallel to find a hot query with a questionable demand for the last. That is, a query having a questionable demand in the search log may be identified first, and then a hot query may be determined using a query having a questionable demand. Alternatively, a hot query may be determined first and Queries with questionable demand may be identified, or queries with questionable demand and hot queries may be identified synchronously before their intersection set is determined.

  To identify whether or not there is a questionable question about a query, the query is subjected to word separation processing based on its meaning, and each word obtained by performing word separation processing is matched in the question attribute database, and the question of each word is identified. The step of determining the trend value, the step of obtaining the question tendency value of the query by combining the question tendency values of each word, and if the question tendency value of the query is larger than a preset question demand threshold, there is a question demand for this query And if not, this query includes the step of determining that there is no questionable demand.

  The question attribute database stores each word obtained through a data discovery method or an artificial arrangement method and a corresponding question tendency value. In the question attribute database, the question tendency value corresponding to each word is determined by factors such as whether or not this word is a question word and the relationship between this word and the question word, but is limited to these elements. Absent. For example, the highest question tendency value is set for the question words such as “what”, “what”, “why”, “how”, “why”, etc. For “method”, “method”, “method”, etc., it is assumed that there is a strong relationship with the question word, and a high question tendency value is set for it. For other words that are weakly related to the question word, small questions A trend value may be set.

  For example, for a query of “fish sashimi bowl” entered by the user, word separation processing based on the meaning is performed to obtain two words of “fish scent of meat” and “method”. Two words are matched in the question attribute database, and it is determined that there is no word that matches "fish savory meat" in the question attribute database, and the question tendency value is assumed to be 0. Match the “method” and determine the question tendency value as 70. By combining the two question tendency values, the question tendency value of the query is determined to be 70, and if the selected question demand threshold is 60 minutes, the query is considered to have question demand.

  The process of determining the hot query will be described in detail in the second embodiment. The last hot query found with questionable demand may be stored as a file in the database. In step 102, a question is formed using the found query and is presented on the page layout of the knowledge and answer platform. In this step, each of the found queries is analyzed and combined based on its meaning to form a question.

  Specifically, the following steps are included. First, a part-of-speech label is attached to a word obtained by performing separation processing based on the meaning of the found query. Then, these words are compared with the grammar of the question sentence set in advance, and the missing words are added to form a question that satisfies the grammar of the question sentence.

  The grammar of the question text only needs to satisfy the grammar requirements of the regular question text, and can be flexibly set. For example, the grammar of the question sentence is set as [adjective / noun + imaginary] + noun + verb + question particle + question sign, and [] is an option. If the word obtained by word separation processing of the query is a noun and a verb, an appropriate question particle and question code are supplemented, and finally the question is combined. After all, using “fish sashimi bowl” as an example, it is separated into words, “fish sashimi bowl” is labeled with a noun, and “method” is labeled with a noun. The question formed by comparing the grammar of the sentence and adding the missing imaginary words, interrogative particles, and symbols may be "What is the way of the fish and meat meat bowl?"

  In addition, for example, if the grammar of the question sentence is noun + verb + question particle + noun + question code, the last question that was formed regarding the query of “fish sashimi” is “ Is there any way to do this? "

  Further, in the conventional knowledge and answer platform, only registered users can submit questions, and the knowledge and answer platform manages registered users using IDs. In order to adapt to this, an ID set of simulated questions can be set in advance, and all IDs in the ID set of simulated questions are acquiesced to the IDs of registered users by the knowledge question-and-answer platform. When a question formed by the method is announced on the page layout of the knowledge question and answer platform, an unused ID is selected from the ID set of mock questions set in advance and announced to simulate a registered user on the knowledge question and answer platform. Submit a question.

  Questions relating to the present invention are not limited to ordinary questions, and may be applied to other types of questions, for example, questions that ask for a file, in which case the relevant knowledge information for this question is: It may be a file uploaded by another user.

  In step 103, related knowledge information of the question is obtained through a page layout on the knowledge question and answer platform. After the question is announced to the knowledge and answer platform, a registered user on the knowledge and answer platform answers the question page layout and provides related knowledge information. Preferably, a good quality answer is determined by the related knowledge information answered in the page layout, and the good quality answer may be determined by the administrator of the knowledge answer platform, and is automatically determined according to the good answer selection strategy that the knowledge answer platform has previously set. May be. The good quality answer selection strategy may be determined by one or any combination of factors such as the grade of the user answering this question, the adoption rate of the question answered by the user, and the length of the related knowledge information.

  Also, after the question is published on the page layout of the knowledge and answer platform, even if the length of the announced time becomes the length of the off-time set in advance, any related information of this question does not appear, or this question If the good answer does not appear, the page layout with this question may be turned off in the knowledge and answer platform. Up to this point, the flow of the first embodiment is completed, and the determination of the hot query will be described below together with the second embodiment.

Second Embodiment FIG. 2 is a flowchart for determining a hot query provided in the second embodiment of the present invention. As shown in FIG. 2, this flow includes the following steps. In step 201, the query is clustered based on relevance to obtain each query set.

  Identifying queries with questionable demand in the search log and determining a hot query in parallel, or first confirming a hot query and later identifying a query with questionable demand in a hot query, The clustering target of this step is a query in the extracted search log.

  When a query having a questionable demand is first identified in the search log and a hot query is confirmed later by a query having a hot request, the clustering target of this step is a query having a questionable demand identified in the search log. After performing clustering, each query included in each query group has high relevance, for example, high relevance between queries such as “World Expo”, “Sebokai”, “Sebo”, etc. And these queries are clustered into a single query.

  The following steps 202 to 203 are executed for each query. In step 202, the search frequency of each query in the query set is combined to determine the search frequency of the entire query set. Based on the search log, the search frequency of each query is statistics, and the search frequency of each query for each query set is combined to reflect the popularity of the overall query set as the search frequency of the overall query set.

  In step 203, it is determined whether or not the search frequency of the query set exceeds a preset hot frequency. If it is exceeded, step 204 is executed, otherwise it is determined that this query set is not a hot query set. . For example, for a query set consisting of “World Expo”, “Sebokai”, and “Sebo”, “World Expo” has a search frequency of 10,000 times within the set time, and “Sebokai” has a search frequency within the set time. Assuming that the search frequency of “Sebo” is 30,000 times within the set time, the search frequency within the set time corresponding to the entire query set is 60,000 times. If the preset hot frequency is 50,000 times, it can be determined that this query set is a hot query set.

  In step 204, this query set is determined to be a hot query set, and one query is selected from the hot query set to be a hot query. The strategy to select one hot query in a hot query group is to select the query with the highest search frequency, select any query, select the query with the highest word meaning completeness, etc. Including. The flow of the second embodiment has been described above. Hereinafter, in the third embodiment, a method for searching related knowledge information based on the method shown in FIG. 1 will be described.

Third Embodiment FIG. 3 is a flowchart of a method for searching related knowledge information provided in the third embodiment of the present invention. As shown in FIG. 3, the method for searching related knowledge information includes the following steps: It is out.

  Step 301 accepts a query entered by a user. In step 302, when a page layout that matches the keyword of the query is searched, and a page layout that matches the keyword of the query is found in the knowledge / answer platform, the page layout that matches the keyword of the query in the knowledge / answer platform. Is included in the search result of the query and is returned to the user.

  According to the flow shown in FIG. 1, if the search engine receives a query entered by the user sent from the browser and then searches the page layout based on the query entered by the user, the background will already follow the flow shown in FIG. Since the user submitted a question and submitted a question to the knowledge and answer platform page layout, when the search engine searches the extracted page layout for a page layout that matches the keyword of this query, It can be matched with a page layout that matches a keyword, and this page layout includes already related questions and related knowledge information of answers to the questions.

  That is, when a user inputs a query with a search engine because a background is found, a hot query having a questioning demand is found in advance, a question is formed, and related knowledge information is acquired from the page layout of the knowledge question and answer platform. Can quickly and correctly feed back related knowledge information in the knowledge question and answer platform in the search result.

  Also, special processing may be applied to the page layout of the knowledge / answer platform, that is, the search engine is allowed to extract the page layout of the good answer existing in the knowledge / answer platform, in other words, on the knowledge / answer platform. If the question submission page layout does not yet have a good answer, the search results fed back to the user will not include the page layout that submitted this question. The above is a detailed description of the method provided by the present invention. Hereinafter, an apparatus for acquiring related knowledge information provided by the present invention in the fourth embodiment will be described in detail.

Fourth Embodiment FIG. 4 is a block diagram of an apparatus for acquiring related knowledge information provided in a fourth embodiment of the present invention. As shown in FIG. 4, the apparatus may include a query discovery unit 400, a question formation unit 410, a question announcement unit 420, and a knowledge acquisition unit 430.

  The query finding means 400 analyzes a search log and finds a hot query having a questionable demand. The search log analyzed by the query discovery unit 400 may be a search log extracted periodically. The question forming unit 410 forms a question using the hot query found by the query finding unit 400. The question announcement means 420 publishes a question on the page layout of the knowledge / answer platform.

  The knowledge acquisition unit 430 acquires related knowledge information of a question using the page layout of the knowledge question / answer platform. The question announcement unit 420 and the knowledge acquisition unit 430 may be a unit that is separated and independent from the knowledge question / answer platform, or may be a unit installed in the knowledge question / answer platform. Specifically, the structure of the query finding unit 400 includes a demand identifying sub-unit 401 and a hot confirmation sub-unit 402, as shown in FIG.

  The demand identifying sub means 401 identifies and outputs a query having a questionable demand from the input query. The hot confirmation sub means 402 confirms and outputs the hot query with the input query. The input of the demand identification sub means 401 may be a query in the extracted search log, and the input of the hot confirmation sub means 402 may be the output of the demand identification sub means 401. In this case, the output of the hot confirmation sub means 402 is a hot query with questionable demand. In this case, the relationship between the demand identifying sub means 401 and the hot confirmation sub means 402 is shown in FIG.

  Alternatively, the input of the hot confirmation sub-means 402 is a query of the search log, and the input of the demand identification sub-means 401 is the output of the hot confirmation sub-means 402. In this case, the output of the demand identification sub-means 401 is a hot with questionable demand. It is a query. In this case, the relationship between the demand identifying sub means 401 and the hot confirmation sub means 402 is shown in FIG.

  Alternatively, the input of the demand identification sub means 401 is a query in the extracted search log, and the input of the hot confirmation sub means 402 is also a query in the search log extracted. In this case, the demand identification sub means 401 and the hot confirmation sub means The relationship with 402 is shown in FIG. In this case, this apparatus may include a sub means for taking the intersection set of the hot confirmation sub means 402 and the demand identification sub means 401, that is, the cross set processing sub means 403 shown in FIG. The output is a hot query with questionable demand.

  Specifically, the structure of the demand identifying sub-unit 401 includes a word separation processing block 601, a block with word points 602, a block with query points 603, and a demand determination block 604, as shown in FIG. . The word separation processing block 601 performs word separation processing based on the meaning of the input query.

  The block with word points 602 matches each word after the word separation processing in the question attribute database, and determines the question tendency value of each word. The question attribute database stores each word obtained by a data discovery method or an artificial arrangement method and a question tendency value corresponding to each word.

  A block 603 with query points combines the question tendency values of each word with each other to obtain a question tendency value of the input query. The demand determination block 604 determines whether or not the question tendency value of the input query exceeds a preset question demand threshold, and if so, determines that there is a question demand for the input query; Confirm that there is no questionable demand for the entered query.

  The question tendency value corresponding to the above word is determined by factors such as whether or not the word is a question word, or the relationship between the word and the question word, but is not limited thereto. Further, the structure of the hot determination sub means 402 specifically includes a clustering processing block 701, a frequency statistics block 702, a hot set determination block 703, and a hot query determination block 704, as shown in FIG. Yes. Clustering processing block 701 performs relevance-based clustering on the query to obtain each query set.

  The frequency statistics block 702 determines the search frequency for each query set by combining the search frequencies of each query for each query set. Since the search frequency for each query set actually represents the search popularity of this query set, the hot set determination block 703 determines a query set whose search frequency exceeds a preset hot frequency as a hot query set.

  The hot query determination block 704 selects one query for each hot query group and sets it as a hot query. The strategies for selecting a hot query in a hot query set are those that select the query with the highest search frequency, those that select any one query, or those that select the query with the highest semantic completeness. However, it is not limited to these.

  As shown in FIG. 4, the question forming unit 410 may specifically include a part-of-speech identification code sub-unit 411 and a sentence combination sub-unit 412. The part-of-speech identification sub means 411 attaches a part of speech label to a word obtained by performing word separation processing based on the meaning of the hot query found by the query finding means 400.

  Here, the part-of-speech identification code sub-means 411 may have a function of word separation processing by itself, that is, the part-of-speech identification code sub-means 411 first applies the hot query found by the query finding means 400 to the hot query. A word separation process based on meaning is performed, and then a part-of-speech label is attached to a word obtained by performing the word separation process. Alternatively, the part of speech identification code sub-unit 411 may not have the function of the word separation processing, and directly uses the result of the word separation processing for this hot query of the word separation processing block 601 in the demand identification sub-unit 401. The part-of-speech label is attached to the word obtained by the word separation processing.

  The sentence combination sub means 412 compares the word obtained by the word separation process with the grammar of the preset question sentence according to the part-of-speech label attached, adds the missing word to the word after the word separation process, Combine questions that fit the grammar of the sentence. In the conventional knowledge and answer platform, only registered users can submit questions, so the knowledge and answer platform manages registered users using IDs.

  In order to adapt to such a situation, an ID set of simulated questions may be set in advance, and an ID in the ID set of simulated questions is silently regarded as a registered user ID by the knowledge question and answer platform. In this case, the question presenting means 420 selects one ID from a preset ID set of simulated questions, imitates the user using the selected ID, and the question formed by the question forming means 410 is used as a knowledge question / answer platform. Announce on the page layout.

  Preferably, the knowledge acquisition unit 430 acquires related knowledge information that the user has answered to the question from the page layout of the knowledge question and answer platform, and determines a good answer from the related knowledge information. The quality answer may be determined by the administrator, and the knowledge / answer platform is determined by one or a combination of the grade of the user who answered the question, the rate of the question answered by the user who answered the question, and the length of the related knowledge information. May be.

  In addition, this apparatus can further improve the quality of the question if any knowledge information related to the question does not appear when the time when the question submitted to the page layout of the knowledge question and answer platform is longer than the preset off time. If the answer has not yet appeared, the question on the knowledge and answer platform includes page layout maintenance means 440 that turns off the page layout on which the question exists, so that a question will not be answered for a long time or a good answer will not appear for a long time, Prevent invalid page layouts.

Fifth Embodiment FIG. 8 is a block diagram of an apparatus for retrieving related knowledge information provided in the fifth embodiment of the present invention. This apparatus includes the apparatus shown in FIG. 4, the user interface means 801, and a page. Layout search means 802. The user interface unit 801 accepts a query input by the user.

  The page layout search unit 802 searches for a page layout that matches the query keyword, and searches for a page layout that matches the query keyword in the page layout in which the device shown in FIG. , The searched page layout is included in the query search results and returned to the user. That is, the page layout extracted by the search engine tapping includes a page layout in which a question in the knowledge question / answer platform exists.

  Also, special processing is performed on the page layout of the knowledge / answer platform, that is, the page layout search means 802 is allowed to search for a page layout with a good answer on the knowledge / answer platform. If the question page layout on the knowledge and answer platform does not yet have a good answer, this question page layout does not exist in the search results returned to the user, that is, the search engine can provide a good answer to the question on the knowledge and answer platform. Install so as not to find page layouts that have not yet appeared.

  The foregoing is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, maintenance, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims (28)

Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Through the page layout of the knowledge questions and answers platform, viewing including the step C to acquire the relevant knowledge information of the question,
Specifically, the step A includes
Identifying queries with questionable demand based on the meaning of words in the search log and determining hot queries with queries with questionable demand; or
Identifying a hot query in the search log and identifying a query with questionable demand based on the meaning of a word in the confirmed hot query, or
Identifying a query having a questionable demand based on the meaning of a word in the search log, and determining a hot query in the search log, and taking a cross set of the identified query having the questionable demand and the confirmed hot query. A method of acquiring related knowledge information characterized by including. The step of identifying a query having the questionable demand specifically includes:
Processing the query into words based on the meaning of the words;
Matching each word obtained by word separation processing in a question attribute database, and determining a question tendency value of each word;
Combining the question tendency values for each word to obtain a query tendency value for the query;
It is determined whether the query tendency value of the query exceeds a preset question demand threshold, and if it exceeds, it is determined that there is question demand for this query, otherwise it is determined that there is no question demand for this query. Including steps and
The related knowledge information according to claim 1 , wherein the question attribute database stores each word obtained through a data discovery method or an artificial arrangement method and a question tendency value corresponding to each word. How to win. The related knowledge information according to claim 2 , wherein the question tendency value corresponding to the word is determined by an element such as whether the word is a question word or a relation between the word and the question word. How to win. The step of determining the hot query specifically includes:
Clustering queries based on relevance to obtain each query set;
Combining the search frequency of each query in each query set to determine the search frequency for each query set;
A query set whose search frequency exceeds a preset hot frequency is determined as a hot query set;
The method of acquiring the related knowledge information as described in any one of Claim 1 to 3 including the step which selects one query from the said hot query group, and makes it a hot query. The step of forming a question using the query found in step A in step B is specifically:
Attaching a part-of-speech label to a word that has undergone word separation processing based on the meaning of the word for the found query;
According to the part-of-speech label attached, the grammar of the question sentence that has been subjected to the word separation process is compared with the grammar of the question sentence that has been set in advance. The method for obtaining related knowledge information according to any one of claims 1 to 4 , further comprising the step of combining matching questions. The step B of publishing the question on the page layout of the knowledge / answer platform specifically includes:
Selecting one ID from a set of preset simulated question IDs, simulating the user using this ID and publishing the question on the page layout of the knowledge and answer platform,
The simulated ID in the set of questions ID, a method of acquiring relevant knowledge information claimed in any one of 5, characterized in that the Mokushi as the ID of the registered user by the knowledge dialogue platform. Specifically, the step C includes
Acquire relevant knowledge information that users on the page layout of the knowledge dialogue platform has answered the above question, according to any one of claims 1 to 5, characterized in that to determine the quality answers from the relevant knowledge information A method to acquire related knowledge information. Step C includes
If the presentation time of the question on the page layout of the knowledge and answer platform is longer than a preset off-time, if any relevant knowledge information of the question has not yet appeared, or a good answer of the question has not yet appeared The method for obtaining related knowledge information according to claim 7 , further comprising: turning off a page layout in which the question exists on the knowledge answer platform. Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Obtaining relevant knowledge information of the question through a page layout of the knowledge question and answer platform;
Including
Step A includes
Identifying queries with questionable demand in the search log and determining hot queries with queries with questionable demand; or
Confirming a hot query in the search log and identifying a query with questionable demand in the confirmed hot query, or
Identifying a query with questionable demand in the search log, and determining a hot query in the search log and taking a cross set of the identified query with questionable demand and the determined hot query,
The step of identifying a query having the questionable demand includes:
Processing the query into words based on the meaning of the words;
Matching each word obtained by word separation processing in a question attribute database, and determining a question tendency value of each word;
Combining the question tendency values for each word to obtain a query tendency value for the query;
It is determined whether the query tendency value of the query exceeds a preset question demand threshold, and if it exceeds, it is determined that there is question demand for this query, otherwise it is determined that there is no question demand for this query. Step and
Including
In the question attribute database, each word obtained through a data discovery method or an artificial arrangement method and a question tendency value corresponding to each word are stored.
A method for acquiring related knowledge information characterized by that . Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Obtaining relevant knowledge information of the question through a page layout of the knowledge question and answer platform;
Including
Step A includes
Identifying queries with questionable demand in the search log and determining hot queries with queries with questionable demand; or
Confirming a hot query in the search log and identifying a query with questionable demand in the confirmed hot query, or
Identifying a query with questionable demand in the search log, and determining a hot query in the search log and taking a cross set of the identified query with questionable demand and the determined hot query,
The step of determining the hot query includes:
Clustering queries based on relevance to obtain each query set;
Combining the search frequency of each query in each query set to determine the search frequency for each query set;
A query set whose search frequency exceeds a preset hot frequency is determined as a hot query set;
Selecting one query from the hot query set to make a hot query;
A method for acquiring related knowledge information characterized by including: Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Obtaining relevant knowledge information of the question through a page layout of the knowledge question and answer platform;
Including
Forming a question using the query found in Step A in Step B includes:
Attaching a part-of-speech label to a word that has undergone word separation processing based on the meaning of the word for the found query;
According to the part-of-speech label attached, the grammar of the question sentence that has been subjected to the word separation process is compared with the grammar of the question sentence that has been set in advance. Combining matching questions and
A method for acquiring related knowledge information characterized by including: Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Obtaining relevant knowledge information of the question through a page layout of the knowledge question and answer platform.
Step B presenting the question on the page layout of the knowledge and answer platform includes:
Selecting one ID from a set of preset simulated question IDs, simulating the user using this ID and publishing the question on the page layout of the knowledge and answer platform,
The IDs in the set of simulated question IDs are silently registered as registered user IDs by the knowledge question and answer platform.
A method for acquiring related knowledge information characterized by that. Analyzing the search log to find a hot search request query with questionable demand;
Step B which forms a question using the query found in Step A and publishes it on the page layout of the knowledge and answer platform;
Obtaining relevant knowledge information of the question through a page layout of the knowledge question and answer platform;
Including
Step C includes
Obtaining relevant knowledge information that the user has answered about the question in the page layout of the knowledge question and answer platform, and determining a good answer from the related knowledge information;
If the presentation time of the question on the page layout of the knowledge and answer platform is longer than a preset off-time, if any relevant knowledge information of the question has not yet appeared, or a good answer of the question has not yet appeared Turning off the page layout on which the question exists on the knowledge and answer platform;
A method for acquiring related knowledge information characterized by including: Accepting the query entered by the user,
Searching for a page layout that matches the keyword of the query,
When a page layout matching the query keyword is obtained by searching the knowledge query platform, the page layout matching the query keyword on the knowledge query platform is included in the query search result and returned to the user. A method for retrieving related knowledge information based on the method for acquiring related knowledge information according to any one of claims 1 to 13 . Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
In page layout of the knowledge questions and answers platform, only contains the knowledge acquisition means to acquire relevant knowledge information of the question,
The query discovery means specifically includes:
A demand identifying sub-means for identifying and outputting a query having a questionable demand based on the meaning of a word in the inputted query;
A hot confirmation sub-means for confirming and outputting a hot query with the input query,
The input of the demand identification sub-means is a query in the search log, the input of the hot confirmation sub-means is the output of the demand identification sub-means, and the output of the hot confirmation sub-means is a hot that has the questionable demand. A query, or
The input of the hot confirmation sub means is a query in the search log, the input of the demand identification sub means is the output of the hot confirmation sub means, and the output of the demand identification sub means is the hot having the questioned demand. A query, or
The input of the demand identification sub means is a query in the search log, and the input of the hot confirmation sub means is also a query in the search log. In this case, the apparatus further includes the hot confirmation sub means the demand identification takes intersection set of sub-unit, to acquire relevant knowledge information you characterized in that it comprises a cross-set processing sub section for outputting a hot queries in doubt demand device. The demand identifying sub-means is specifically:
A word separation processing block that separates the query into words based on the meaning of the word;
A block with word points that matches each word obtained by the word separation processing in the question attribute database and determines the question tendency value of each word,
A block with a query point that obtains the question tendency value of the input query by combining the question tendency values of the words,
It is determined whether the question tendency value of the input query exceeds a preset question demand threshold, and if it is determined that the query demand threshold is exceeded, it is determined that there is a question demand for the input query, and A demand determination block that determines that there is no questionable demand for the input query if it is determined that the questionable demand threshold is not exceeded,
The related knowledge information according to claim 15 , wherein the question attribute database stores each word obtained through a data discovery method or an artificial arrangement method and a question tendency value corresponding to each word. The device to acquire. The related knowledge according to claim 16 , wherein the question tendency value corresponding to each word is determined by factors such as whether or not the word is a question word, or a relation between the word and the question word. A device that acquires information. A clustering processing block that performs query-based clustering for each query to obtain each query set;
A frequency statistics block for determining the search frequency for each query set by combining the search frequencies of each query for each query set;
A hot set confirmation block for determining a query set whose search frequency exceeds a preset hot frequency as a hot query set;
The apparatus for acquiring related knowledge information according to any one of claims 15 to 17 , further comprising: a hot query determination block that selects one query from each hot query set and sets it as a hot query. Specifically, the question forming means includes:
A part-of-speech identification sub-means for attaching a part-of-speech label to a word that has been subjected to word separation processing based on the meaning of the word for the query found by the query discovery means
According to the part-of-speech label attached, the grammar of the question sentence that has been subjected to the word separation process is compared with the grammar of the question sentence that has been set in advance, and the word that is missing from the word that has undergone the word separation process is added. The apparatus for acquiring related knowledge information according to any one of claims 15 to 18 , further comprising: a sentence combination sub-unit for combining questions to be asked. Specifically, the question announcement means is:
Selecting one ID from a set of preset simulated question IDs, simulating a user using the selected ID and publishing the question formed by the question forming means on the page layout of the knowledge question and answer platform;
The apparatus for acquiring related knowledge information according to any one of claims 15 to 19 , wherein an ID in the set of simulated question IDs is silently registered as an ID of a registered user by the knowledge question and answer platform. Specifically, the knowledge acquisition means includes:
The knowledge acquired relevant knowledge information dialogue platform page layout users to answer for the question, according to any one of the relevant knowledge and wherein the determining the quality answer from information claims 15 to 20 A device that acquires related knowledge information. The knowledge acquisition means includes
When the announcement time on the page layout of the question / answer platform of the question is longer than a preset off time, any related knowledge information of the question has not yet appeared, or a good answer of the question has not yet appeared The apparatus for acquiring related knowledge information according to claim 21 , further comprising page layout maintenance means for turning off a page layout on which the question exists on the knowledge question and answer platform. Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
Knowledge acquisition means for acquiring related knowledge information of the question in the page layout of the knowledge question and answer platform,
The query discovery means includes
A demand identification sub-means for identifying and outputting a query having a questionable demand in the input query;
A hot confirmation sub-means for confirming and outputting a hot query with the input query,
The input of the demand identification sub-means is a query in the search log, the input of the hot confirmation sub-means is the output of the demand identification sub-means, and the output of the hot confirmation sub-means is a hot that has the questionable demand. A query, or
The input of the hot confirmation sub means is a query in the search log, the input of the demand identification sub means is the output of the hot confirmation sub means, and the output of the demand identification sub means is the hot having the questioned demand. A query, or
The input of the demand identification sub means is a query in the search log, and the input of the hot confirmation sub means is also a query in the search log. In this case, the apparatus further includes the hot confirmation sub means Taking a cross set of the demand identifying sub means, and including a cross set processing sub means for outputting a hot query having a questionable demand,
The demand identifying sub-means
A word separation processing block that separates the query into words based on the meaning of the word;
A block with word points that matches each word obtained by the word separation processing in the question attribute database and determines the question tendency value of each word,
A block with a query point that obtains the question tendency value of the input query by combining the question tendency values of the words,
It is determined whether the question tendency value of the input query exceeds a preset question demand threshold, and if it is determined that the query demand threshold is exceeded, it is determined that there is a question demand for the input query, and A demand determination block that determines that there is no questionable demand for the input query if it is determined that the questionable demand threshold is not exceeded,
In the question attribute database, each word obtained through a data discovery method or an artificial arrangement method and a question tendency value corresponding to each word are stored.
The apparatus which acquires the related knowledge information characterized by this. Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
Knowledge acquisition means for acquiring related knowledge information of the question in the page layout of the knowledge question and answer platform,
The query discovery means includes
A demand identification sub-means for identifying and outputting a query having a questionable demand in the input query;
A hot confirmation sub-means for confirming and outputting a hot query with the input query,
The input of the demand identification sub-means is a query in the search log, the input of the hot confirmation sub-means is the output of the demand identification sub-means, and the output of the hot confirmation sub-means is a hot that has the questionable demand. A query, or
The input of the hot confirmation sub means is a query in the search log, the input of the demand identification sub means is the output of the hot confirmation sub means, and the output of the demand identification sub means is the hot having the questioned demand. A query, or
The input of the demand identification sub means is a query in the search log, and the input of the hot confirmation sub means is also a query in the search log. In this case, the apparatus further includes the hot confirmation sub means Taking a cross set of the demand identifying sub means, and including a cross set processing sub means for outputting a hot query having a questionable demand,
A clustering processing block that performs query-based clustering for each query to obtain each query set;
A frequency statistics block for determining the search frequency for each query set by combining the search frequencies of each query for each query set;
A hot set confirmation block for determining a query set whose search frequency exceeds a preset hot frequency as a hot query set;
A hot query determination block that selects one query from each hot query set and sets it as a hot query;
The apparatus which acquires the related knowledge information characterized by including this. Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
Knowledge acquisition means for acquiring related knowledge information of the question in the page layout of the knowledge question and answer platform,
The question forming means
A part-of-speech identification sub-means for attaching a part-of-speech label to a word that has been subjected to word separation processing based on the meaning of the word for the query found by the query discovery means;
According to the part-of-speech label attached, the grammar of the question sentence that has been subjected to the word separation process is compared with the grammar of the question sentence that has been set in advance, and the word that is missing from the word that has undergone the word separation process is added. Sentence combination sub-means to combine questions
The apparatus which acquires the related knowledge information characterized by including this. Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
Knowledge acquisition means for acquiring related knowledge information of the question in a page layout of the knowledge question-and-answer platform;
Including
The question announcement means is:
Selecting one ID from a set of preset simulated question IDs, simulating a user using the selected ID and publishing the question formed by the question forming means on the page layout of the knowledge question and answer platform;
The IDs in the set of simulated question IDs are silently registered as registered user IDs by the knowledge question and answer platform.
The apparatus which acquires the related knowledge information characterized by this. Query discovery means to analyze search logs and find hot queries with questionable demand,
Question forming means for forming a question using the hot query found by the query finding means;
Question announcement means for announcing the question on the page layout of the knowledge and answer platform;
Knowledge acquisition means for acquiring related knowledge information of the question in a page layout of the knowledge question-and-answer platform;
Including
The knowledge acquisition means includes
Means for acquiring relevant knowledge information for a user to answer the question in the page layout of the knowledge question and answer platform, and determining a good answer from the related knowledge information;
When the announcement time on the page layout of the question / answer platform of the question is longer than a preset off time, any related knowledge information of the question has not yet appeared, or a good answer of the question has not yet appeared A page layout maintenance means for turning off the page layout on which the question exists on the knowledge and answer platform;
The apparatus which acquires the related knowledge information characterized by including this. An apparatus for acquiring related knowledge information according to any one of claims 15 to 27 ;
A user interface means to accept queries entered by the user;
28. A page layout that searches for a page layout that matches a keyword of the query and obtains related knowledge information according to any one of claims 15 to 27 , wherein the query keyword is a page layout in which a question in a knowledge question and answer platform is announced. And a page layout search means for causing the user to return the searched page layout to the search result of the query when a page layout that matches is searched. An apparatus for searching related knowledge information.

Images