WO2017088497A1 - 一种实体推荐方法、装置、设备及计算机存储介质 - Google Patents
一种实体推荐方法、装置、设备及计算机存储介质 Download PDFInfo
- Publication number
- WO2017088497A1 WO2017088497A1 PCT/CN2016/089649 CN2016089649W WO2017088497A1 WO 2017088497 A1 WO2017088497 A1 WO 2017088497A1 CN 2016089649 W CN2016089649 W CN 2016089649W WO 2017088497 A1 WO2017088497 A1 WO 2017088497A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- entity
- entities
- description information
- recommended
- clicked
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Definitions
- the present invention relates to the field of Internet search technologies, and in particular, to an entity recommendation method and apparatus.
- the search engine searches according to the query word, and outputs a number of search result pages matching the query word to the user.
- a recommendation card may be inserted in the displayed search result page for recommending a plurality of recommendation entities related to the query term and having a single category, and the user may click on an entity in the recommendation card to jump from the current search result page. Go to the search results page that matches the entity you clicked on.
- the recommended entities currently output are often related entities with a single category, and the correlation between the recommended entities and the query terms is not clear. Therefore, the user needs to manually filter and find the required entities in the recommended entities, resulting in the prior art.
- the category of the recommended entity is single, and the efficiency of obtaining the recommended entity from the search results page is relatively low.
- an embodiment of the present invention provides an entity recommendation method and apparatus, which improves the diversity of recommended entities in a search result page, and improves the efficiency of obtaining a recommendation entity from a search result page.
- An aspect of an embodiment of the present invention provides an entity recommendation method, including:
- the description information corresponding to each entity set is displayed in the search result page matching the query word, and each recommended entity included in the entity set corresponding to the clicked description information is displayed according to the description information clicked by the user.
- the obtained query word obtains at least two entity sets, including:
- the at least two entity sets are obtained from at least two candidate sets.
- any possible implementation manner further provide an implementation manner of displaying description information corresponding to each entity set in a search result page that matches the query term, including:
- the corresponding display of the description information corresponding to each entity set and the target image is performed.
- any possible implementation manner further provide an implementation manner of obtaining a target image that matches each entity set, including:
- At least two candidate pictures that match the set of entities are obtained, and the at least two candidate pictures are puzzled to generate a target picture that matches the set of entities.
- any of the possible implementations further provide an implementation manner of displaying, according to the description information that the user clicks, the recommended entities included in the entity set corresponding to the clicked description information, including:
- the page rendering is performed by using each recommended entity included in the entity set corresponding to the description information clicked by the user to display a new page including each recommended entity.
- the above-mentioned aspect and any possible implementation manner further provide an implementation manner in which the recommended entities are displayed by using a banner in the new page; the method further includes:
- the search result matching the clicked recommended entity is displayed below the banner in the new page according to the recommended entity clicked in each recommended entity displayed by the user in the through-column.
- the navigation information includes the query word, description information corresponding to the set of entities in which the recommended entity is located, and the recommended entity that is clicked.
- any of the possible implementations further provide an implementation manner of displaying, according to the description information that the user clicks, the recommended entities included in the entity set corresponding to the clicked description information, including:
- a search result page matching the query word The drop-down box is displayed below the description information, and the recommended entities included in the entity set corresponding to the description information that the user clicks are displayed by using the drop-down box.
- An aspect of an embodiment of the present invention provides an entity recommendation apparatus, including:
- An entity obtaining module configured to obtain at least two entity sets according to the obtained query words
- An information obtaining module configured to generate corresponding description information for each entity set according to a common feature of each entity set and the query word;
- a recommendation display module configured to display description information corresponding to each entity set in a search result page matching the query word, and display, according to the description information clicked by the user, the entity set corresponding to the clicked description information Each recommended entity.
- the at least two entity sets are obtained from at least two candidate sets.
- the apparatus further includes: a picture generating module, configured to obtain a target picture that matches each entity set;
- the recommendation display module is configured to: perform corresponding display of the description information corresponding to each entity set and the target image in a search result page that matches the query word.
- At least two candidate pictures that match the set of entities are obtained, and the at least two candidate pictures are puzzled to generate a target picture that matches the set of entities.
- recommendation display module is specifically configured to:
- the page rendering is performed by using each recommended entity included in the entity set corresponding to the description information clicked by the user to display a new page including each recommended entity.
- each recommended entity is displayed by using a banner in the new page; the recommended display module is further configured to:
- the search result matching the clicked recommended entity is displayed below the banner in the new page according to the recommended entity clicked in each recommended entity displayed by the user in the through-column.
- recommendation display module is further configured to: display navigation between the clicked recommendation entity and the query word in the pass column information;
- the navigation information includes the query word, description information corresponding to the set of entities in which the recommended entity is located, and the recommended entity that is clicked.
- recommendation display module is specifically configured to:
- a search result page matching the query word The drop-down box is displayed below the description information, and the recommended entities included in the entity set corresponding to the description information that the user clicks are displayed by using the drop-down box.
- the technical solution provided by the embodiment of the present invention can provide description information of multiple entity sets in a search result page, and can further display each recommended entity included in the entity set, and can only provide a recommendation entity with a single category in the prior art.
- the technical solution provided by the embodiment of the present invention can output a multi-category and a large number of recommended entities, thereby improving the diversity of recommended entities in the search result page.
- the recommendation relationship between the query word and the recommendation entity may be output by outputting the description information corresponding to the entity set. To help users quickly get the entities they need and improve the efficiency of getting recommended entities from search results pages.
- FIG. 1 is a schematic flowchart of an entity recommendation method according to an embodiment of the present invention.
- 2(a) to 2(h) are diagrams showing an example of a combination manner when at least two candidate pictures are generated by a target image according to an embodiment of the present invention
- FIG. 3 is a first exemplary diagram of a recommended entity in a search result page in the prior art
- 4(a) to 4(c) are diagrams showing a first example of a recommended entity in a search result page according to an embodiment of the present invention
- 5(a) to 5(b) are diagrams showing a second example of recommending entities in a search result page according to an embodiment of the present invention.
- FIG. 6 is a second exemplary diagram of a recommended entity in a search result page in the prior art
- FIG. 7(a)-7(c) are third exemplary diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- 8(a) to 8(b) are fourth exemplary diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- FIG. 9 is a functional block diagram of an entity recommendation apparatus according to an embodiment of the present invention.
- the word “if” as used herein can be interpreted as “in... “or” or “in response to the determination” or “in response to the detection.”
- the phrase “if determined” or “if detected (conditions or events stated)” may be interpreted Becoming “when determined” or “in response to determination” or “when detecting (stated condition or event)” or “in response to detection (condition or event stated)”.
- FIG. 1 it is a schematic flowchart of an entity recommendation method according to an embodiment of the present invention. As shown in the figure, the method includes the following steps:
- the query word input by the user may be received, and then searched according to the query word to obtain a plurality of search results that match the query word, and finally, the search result page is used to display a plurality of search results to the user.
- the generated search results page you need to provide users with multi-dimensional, multi-category recommended entities at the same time.
- at least two entity sets are obtained according to the obtained query word.
- the method for obtaining at least two entity sets according to the obtained query words may include, but is not limited to, the following two types:
- a first type obtaining at least two entities with the highest heat among the entities related to the search entity according to the search entity included in the query word; and clustering the at least two entities to obtain a Describe at least two sets of entities.
- the query term only includes the search entity, or the query term is a search entity, it indicates that the user's search demand information cannot be determined according to the query word, and therefore the user's search demand is not clear, and may be related to the search entity.
- the highest heat among all entities At least two entities.
- the clustering algorithm may be used to perform clustering operations on at least two entities with the highest heat, and the at least two entities are divided into at least two entity sets, thereby obtaining at least two entity sets according to the obtained query words.
- a second type obtaining, according to the search entity and the search requirement information included in the query word, at least two candidate sets that match the search entity and the search requirement information, according to each candidate set and the search demand information Correlation, the at least two entity sets are obtained from at least two candidate sets.
- the query term includes the search entity and the search demand information
- the user needs to determine the search demand information according to the query word. Therefore, if the user's search demand is clear, the search entity and the search demand information can be obtained according to the search requirement. At least two candidate sets with the highest information relevance.
- each candidate set may be sorted according to the relevance of each candidate set and the search demand information, and the order of the related sets is obtained in order of large to small, to obtain a sort result, and finally obtained from the sort result.
- the highest ranked at least two candidate sets are used as the at least two entity sets.
- IQ is an impetuous dog, "a dog that looks like a wolf like a husky", and a "no-attack medium-sized dog with a gentle character like a huskies.”
- the user's search demand is very clear, that is, interested in the feeding mode of the Husky and other related pets, and the highest degree of relevance to the user's search demand information can be obtained.
- Entity collection such as "as with huskies Medium-sized dogs that are difficult to raise, "small pet dogs that are suitable for domestication with Husky,” and "pets that are as loyal and well-trained as Huskies.”
- the different search requirements of the query words input by the user are divided, and the entities that are interested in the search requirements are sorted according to the categories, forming different sets of entities, and finally recommending the set of entities to The user is used to satisfy different search requirements of the user, wherein the order of each entity set may be based on the correlation between the entity set and the query word input by the user for the search requirement, and the entity with the highest relevance to the search requirement of the user. In the top position.
- the common feature of each entity set and the query word may be extracted, and then corresponding descriptions are generated for each entity set according to the common features of each entity set and the query words. information.
- the query word is “Beethoven”
- the entity collection includes recommended entities, such as “Hawking” and “Vincent Van Gogh”.
- the common feature of the entity collection and query words is “Celebrity of the body”
- the description information corresponding to the entity set may be "a celebrity like Beethoven”.
- the description information corresponding to the entity set not only reflects the common feature between the entity set and the search term, but also reflects the reason for recommending the entity set to the user, so that the user can better understand the recommended entities included in the entity set. And better understanding the relationship between the recommended entity set and the query word searched by the user, or the relationship between the entity set and the user's search intent, thereby enabling the user to quickly select the entity of interest or need, In turn, the efficiency of the user to obtain the recommended entities required is improved.
- the description information corresponding to each entity set may be displayed in a search result page that matches the query word.
- the description information corresponding to each entity set may be displayed in the middle of any two search results in the search result page matching the query word.
- a target picture matching the set of entities may also be obtained; and then, in the search result page matching the query word, the description information corresponding to each entity set and the target picture are performed. Corresponding display.
- the description information corresponding to each entity set is horizontally displayed, and the target image matching the entity set corresponding to the description information is displayed at a position below each description information displayed. .
- the description information corresponding to each entity set is associated with the corresponding target image by the corresponding display of the description information corresponding to the target set, so that the user can directly perceive the recommended entity set through the target image.
- the method for obtaining a target image that matches each entity set may include, but is not limited to, the following two types:
- a second type for each of the entity sets, obtaining at least two candidate pictures that match the set of entities, and jigating the at least two candidate pictures to generate a target picture that matches the set of entities .
- multiple graphs matching the entity set may be aggregated, such as three candidate pictures, four candidate pictures, or an aggregation of five candidate pictures, or even more representative. Aggregation of pictures of a single entity.
- a picture corresponding to each recommended entity in the first plurality of recommended entities in the entity set may be selected, and then the pictures are arranged and aggregated.
- a picture in the entity set that best reflects the search requirement represented by the entity set may be selected, and similar pictures are aggregated on the visual, style, color, and the like of the picture to generate a target picture.
- FIG. 2( a ) to FIG. 2( h ) are exemplary diagrams of a combination manner when generating a target picture for at least two candidate pictures according to an embodiment of the present invention, as shown in FIG. 2( a ) to FIG. 2( f ).
- at least two candidate pictures may be a plurality of candidate picture combinations to generate a target picture.
- at least two candidate pictures may be shadow-added to generate a target picture.
- FIG. 2(h) at least two candidate pictures may be combined in 3D to generate a target picture.
- the manner in which the target picture is generated by using at least two candidate pictures is only an example. The embodiment of the present invention does not specifically limit the generation manner.
- the description information corresponding to each entity set is displayed in the search result page that matches the query word, if there is a set of entities that the user needs to know in each entity set, The user can use the gesture to slide left or right to the right in the description information corresponding to each entity set, so that the user can browse to the real Description of the body collection.
- the user clicks the description information of the entity collection so that the recommended entities included in the entity set corresponding to the clicked description information may be further displayed according to the description information that the user clicks, so that Providing the recommended entities included in the entity collection to the user.
- the method for displaying each recommended entity included in the entity set corresponding to the clicked description information may include, but is not limited to, the following two types:
- the first type performing page rendering by using each recommended entity included in the entity set corresponding to the description information clicked by the user to display a new page including each recommended entity.
- each recommended entity may be displayed in the new page by using a banner; and then, according to the recommended entity clicked by each recommended entity displayed by the user in the portal, in the new page. Below the banner, search results that match the recommended entity being clicked are displayed.
- a new search may be performed according to the entity set corresponding to the clicked description information, and a banner is displayed at the top position of the new page.
- the banner can be fixed at the top of the search result page, so that the user can quickly browse all the recommended entities included in the entity set corresponding to the clicked description information in the portal.
- the user can further click on the required recommendation entity, and according to the recommended entity clicked by the user, the search result matching the clicked recommended entity can be automatically rendered under the banner in the new page by the asynchronous loading technology, wherein the recommended entity is clicked.
- the search results that match the recommended entities that are clicked are all performed in the new page, and do not need to jump to other pages, so there is no need to switch back and forth between the pages, so that it is very convenient to click on the
- the required recommendation entities and quick access to search results matching the recommended entities of the click greatly improve the efficiency of obtaining multiple recommended entities and reduce the search time.
- the navigation information between the recommended entity and the query word may be displayed in the navigation bar, where the navigation information includes the query word and a description corresponding to the set of entities in which the recommended entity is located. Information and the recommended entity that was clicked.
- the banner can be composed of two parts: navigation information and each recommended entity.
- the navigation information can be displayed above the recommended entities in the portal bar, so that the user can quickly understand the recommendation logic between the recommended entities from the query word to the click, so that the user can quickly locate the recommended entity that is clicked.
- the query word in the navigation information is located on the left side
- the description information corresponding to the set of entities in which the recommended entity is located is located in the middle
- the recommended entity that is clicked is located on the right side.
- the navigation information can be "Beethoven” and the same celebrity as Beethoven's celebrity > Vincent Van Gogh, where "Beethoven” is the query word entered by the user, "like Beethoven's body and mind "Celebrity” is the description information corresponding to the set of entities clicked by the user in each entity collection, and "Vincent Van Gogh” is a recommendation information that the user further clicks in each recommendation information included in the entity collection.
- the description information of the entity set may be refined, and the refined information is used as the description information in the navigation information.
- the navigation information may also be “Beethoven> a similar celebrity of the disabled” Vincent ⁇ Van Gogh, which reduces information redundancy and improves the efficiency of information acquisition.
- the name of each recommended entity and the picture corresponding to each recommended entity may be displayed in a horizontal direction.
- the recommended entities may be sorted according to the relevance of each recommended entity and the search demand information, and the higher the relevance is The more the body is in the front position, the user can slide left or right to navigate through the recommended entities.
- the search result page matching the query word may be returned by the above new page including each recommended entity, and other entity sets may be further selected.
- the second type displaying a drop-down box below the description information in the search result page matching the query word according to the description information clicked by the user, and displaying the description information clicked by the user by using the drop-down box Each recommended entity contained in the entity collection.
- the left swipe gesture or the right swipe gesture may be used to move each entity set, and then browse to each entity set.
- the drop-down box may be displayed below the description information corresponding to each entity set in the search result page matching the query word, and then the user clicks on the drop-down box.
- Each recommended entity included in the entity set corresponding to the description information for example, the entity set may include 12 to 20 recommended entities, and the recommended entities may be sorted according to the relevance of the recommended entity and the search requirement, and the correlation is higher. The recommended entity is ranked higher.
- the user can also use the left swipe gesture or the right swipe gesture to move each recommended entity to browse to all recommended entities included in the entity set, so that the user can browse in the search result page.
- the collection of entities in each dimension, as well as multiple recommended entities in the collection of entities, improves the efficiency of obtaining recommended entities.
- Beethoven a collection of classical music styles, a poor, hard-working music artist
- other search needs that the user may be interested in when the main search needs are met Including: a celebrity like Beethoven, a poor artist like Beethoven, and a musician with the same style as Beethoven. Wait.
- FIG. 3 is a first example diagram of a recommended entity in a search result page in the prior art.
- the recommended entity recommends “Mozart”, “Chopin” and “Bybon” around Beethoven. Schubert, "Bach” and other entities, these are well-known musicians in the world, the dimension of the recommended entity is single and the description of the "related person” is weak to the recommended entity, for the user in "and Beethoven”
- the two search needs of artists who are born in poverty and "like celebrities like Beethoven” cannot be satisfied.
- FIG. 4( a ) to FIG. 4( c ) are first example diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- entities such as Haydn, Mozart, Chopin, and Bach are grouped together into a collection of entities, extracting common features and generating descriptive information “the same creative style as Beethoven’s.
- FIG. 5( a ) to FIG. 5( b ) are second exemplary diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- Fig. 5(a) if the user is interested in "a musician who has the same creative style as Beethoven", the description information is clicked and displayed below each entity model displayed in the search result page related to the query word.
- the recommended entities in the entity set corresponding to the description information are displayed in the drop-down box, such as "Mozart", “Chopin”, "Hayton” and "Bach”.
- the user can browse to all recommended entities by using a swipe left gesture or a swipe right gesture on each recommendation entity.
- Fig. 5(b) if the user is interested in "a celebrity like Beethoven,” he or she clicks on the description information and displays it below each entity model displayed in the search result page related to the query word.
- the recommended entities in the entity set corresponding to the description information are displayed in the drop-down box, such as "Van Gogh”, “Hawkin”, “Shi Tiesheng” and "A Bing".
- FIG. 6 is a second example diagram of the recommended entity in the search result page in the prior art.
- the recommended entity recommends “Shawshank Redemption” and “Chasing” around the Forrest Gump Dress.
- There are several entities such as “Dream Heart” and “Becoming a Person”. These are the famous movies of Oscar.
- the dimension of the recommendation entity is single and the description of the "relevant recommendation” is weak to the recommendation entity.
- FIG. 7( a ) to FIG. 7( c ) are third exemplary diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- the entities such as "Birdman”, “Flying over the madhouse”, “the burning years”, and “the fantasy drifting of the juveniles” are summarized in a collection of entities, extracting common features and generating description information.
- the Oscar-winning work of the box office is as high as that of Forrest Gump."
- the collection of entities such as “Shawshank's Redemption”, “When Happiness Comes to Knock”, “Never Give Up”, “Spring of the Cowboy Class”, etc., and adds descriptive information to the collection of the entity "as classic as Forrest Gump Must-see inspirational movie.”
- you can add a collection of entities such as "Mermaid”, “Captain Philip”, “Philadelphia Story” and “The Rest of the Desert”, and add a description for the collection of the entity "The same popular movie starring Tom Hanks” .
- the three entity sets are respectively matched with the multi-image aggregated picture, and the pictures of the five most representative entities in each entity set are combined, compared with the conventional technical solution.
- the technical solution provided by the embodiment of the invention achieves the upgrade from the entity recommendation to the entity collection recommendation.
- FIG. 8( a ) to FIG. 8( b ) are fourth exemplary diagrams of recommended entities in a search result page according to an embodiment of the present invention.
- Fig. 8(a) if the user is interested in "inspirational movies that are as classic as Forrest Gump", click on the description information to display the entity models in the search results page related to the query words.
- the drop-down box is displayed below, and the recommended entities in the entity set corresponding to the description information are displayed in the drop-down box, such as "never compromise", “Jordan passer”, “successor” and "weakness”.
- Embodiments of the present invention further provide an apparatus embodiment for implementing the steps and methods in the foregoing method embodiments.
- FIG. 9 is a functional block diagram of an entity recommendation apparatus according to an embodiment of the present invention. As shown, the device includes:
- the entity obtaining module 91 is configured to obtain at least two entity sets according to the obtained query words
- the information obtaining module 92 is configured to: according to common features of each entity set and the query term, Generating corresponding description information for each entity set;
- a recommendation display module 93 configured to display description information corresponding to each entity set in a search result page matching the query word, and display an entity collection corresponding to the clicked description information according to the description information clicked by the user Each recommended entity is included.
- the entity obtaining module 91 is specifically configured to:
- the at least two entity sets are obtained from at least two candidate sets.
- the apparatus may further include: a picture generating module 94, configured to obtain a target picture that matches each entity set;
- the recommendation display module 93 is configured to: perform a corresponding display of the description information corresponding to each entity set and the target image in a search result page that matches the query word.
- the image generating module 94 is specifically configured to:
- At least two candidate pictures that match the set of entities are obtained, and the at least two candidate pictures are puzzled to generate a target picture that matches the set of entities.
- the recommendation display module 93 is specifically configured to:
- the page rendering is performed by using each recommended entity included in the entity set corresponding to the description information clicked by the user to display a new page including each recommended entity.
- each of the recommended entities is displayed in the new page by using a banner; the recommendation display module 93 is further configured to:
- the search result matching the clicked recommended entity is displayed below the banner in the new page according to the recommended entity clicked in each recommended entity displayed by the user in the through-column.
- the recommendation display module 93 is further configured to: display navigation information between the clicked recommendation entity and the query word in the pass column;
- the navigation information includes the query word, description information corresponding to the set of entities in which the recommended entity is located, and the recommended entity that is clicked.
- the recommendation display module 93 is specifically configured to:
- a drop-down box is displayed below the description information in the search result page matching the query word, and displaying, by using the drop-down box, the entity set corresponding to the description information that the user clicks includes Each recommended entity.
- At least two entity sets are obtained according to the obtained query words; thus, corresponding description information is generated for each entity set according to the common features of each entity set and the query words;
- the search result page corresponding to the query word displays the description information corresponding to each entity set, and displays each recommended entity included in the entity set corresponding to the clicked description information according to the description information clicked by the user.
- the technical solution provided by the embodiment of the present invention can provide multiple entity sets in a search result page.
- the description information of the combination and can further display the recommended entities included in the entity set, and the technical solution provided by the embodiment of the present invention can output multiple categories as compared with the technical solution that only provides the recommendation entity with a single category in the prior art. A large number of recommended entities, which improves the diversity of recommended entities in the search results page.
- the recommendation relationship between the query word and the recommendation entity may be output by outputting the description information corresponding to the entity set. To help users quickly get the entities they need and improve the efficiency of getting recommended entities from search results pages.
- the disclosed system, apparatus, and method may be implemented in other manners.
- the device embodiments described above are merely illustrative.
- the division of the unit is only a logical function division.
- multiple units or components may be combined. Or it can be integrated into another system, or some features can be ignored or not executed.
- the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
- the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
- each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
- the above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional units.
- the above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium.
- the above software functional unit is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps.
- the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. .
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
一种实体推荐方法、装置、设备及计算机存储介质。方法包括根据获得的查询词,获得至少两个实体集合(S101);从而,根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息(S102);进而,在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息(S103),以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体(S104)。上述技术方案能够在搜索结果页中提供多个实体集合的描述信息,并能够进一步显示实体集合所包含的各推荐实体,提高了搜索结果页中推荐实体的多样性,以及提高了从搜索结果页中获取推荐实体的效率。
Description
本申请要求了申请日为2015年11月25日,申请号为201510828555.X发明名称为“一种实体推荐方法及装置”的中国专利申请的优先权。
本发明涉及互联网搜索技术领域,尤其涉及一种实体推荐方法及装置。
目前,用户在输入查询词(Query)后,搜索引擎根据该查询词进行搜索,并向用户输出若干与该查询词相匹配的搜索结果页。现有技术中,在显示的搜索结果页中可以***一张推荐卡片,用于推荐与查询词相关且类别单一的若干推荐实体,用户可以在推荐卡片中点击一个实体,从当前搜索结果页跳转到与该点击的实体相匹配的搜索结果页。
然而,目前输出的推荐实体往往是类别单一的相关实体,而且推荐实体与查询词之间的相关关系不明确,因此需要用户在推荐实体中进行人工筛选和查找所需要的实体,导致现有技术中推荐实体的类别单一,且从搜索结果页中获取推荐实体的效率比较低。
发明内容
有鉴于此,本发明实施例提供了一种实体推荐方法及装置,提高了搜索结果页中推荐实体的多样性,以及提高了从搜索结果页中获取推荐实体的效率。
本发明实施例的一方面,提供一种实体推荐方法,包括:
根据获得的查询词,获得至少两个实体集合;
根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;
在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述根据获得的查询词,获得至少两个实体集合,包括:
根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合;或者,
根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,包括:
获得与各实体集合相匹配的目标图片;
在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,获得与各实体集合相匹配的目标图片,包括:
对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片;或者,
对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体,包括:
利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述新页面中利用通栏显示各推荐实体;所述方法还包括:
根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述方法还包括:
在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;
其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体,包括:
根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页
中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
本发明实施例的一方面,提供一种实体推荐装置,包括:
实体获取模块,用于根据获得的查询词,获得至少两个实体集合;
信息获取模块,用于根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;
推荐显示模块,用于在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述实体获取模块,具体用于:
根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合;或者,
根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述装置还包括:图片生成模块,用于获得与各实体集合相匹配的目标图片;
所述推荐显示模块,具体用于:在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述图片生成模块,具体用于:
对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片;或者,
对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述推荐显示模块,具体用于:
利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述新页面中利用通栏显示各推荐实体;所述推荐显示模块,还用于:
根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述推荐显示模块,还用于:在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;
其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
如上所述的方面和任一可能的实现方式,进一步提供一种实现方式,所述推荐显示模块,具体用于:
根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页
中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
由以上技术方案可以看出,本发明实施例具有以下有益效果:
本发明实施例提供的技术方案能够在搜索结果页中提供多个实体集合的描述信息,并且能够进一步显示实体集合所包含的各推荐实体,与现有技术中只能提供类别单一的推荐实体的技术方案相比,本发明实施例所提供的技术方案能够输出多类别、数量多的推荐实体,从而提高了搜索结果页中推荐实体的多样性。而且,与现有技术中推荐实体与查询词之间推荐关系不明确的技术方案相比,本发明实施例中通过输出实体集合对应的描述信息,可以输出查询词与推荐实体之间的推荐关系,帮助用户快速获得所需要的实体,提高了从搜索结果页中获取推荐实体的效率。
为了更清楚地说明本发明实施例的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其它的附图。
图1是本发明实施例所提供的实体推荐方法的流程示意图;
图2(a)~图2(h)是本发明实施例所提供的至少两个候选图片生成目标图片时的组合方式示例图;
图3是现有技术中在搜索结果页中推荐实体的第一示例图;
图4(a)~图4(c)是本发明实施例所提供的在搜索结果页中推荐实体的第一示例图;
图5(a)~图5(b)是本发明实施例所提供的在搜索结果页中推荐实体的第二示例图;
图6是现有技术中在搜索结果页中推荐实体的第二示例图;
图7(a)~图7(c)是本发明实施例所提供的在搜索结果页中推荐实体的第三示例图;
图8(a)~图8(b)是本发明实施例所提供的在搜索结果页中推荐实体的第四示例图;
图9是本发明实施例所提供的实体推荐装置的功能方块图。
为了使本发明的目的、技术方案和优点更加清楚,下面结合附图和具体实施例对本发明进行详细描述。
应当明确,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其它实施例,都属于本发明保护的范围。
在本发明实施例中使用的术语是仅仅出于描述特定实施例的目的,而非旨在限制本发明。在本发明实施例和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式,除非上下文清楚地表示其他含义。
应当理解,本文中使用的术语“和/或”仅仅是一种描述关联对象的关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中字符“/”,一般表示前后关联对象是一种“或”的关系。
取决于语境,如在此所使用的词语“如果”可以被解释成为“在……
时”或“当……时”或“响应于确定”或“响应于检测”。类似地,取决于语境,短语“如果确定”或“如果检测(陈述的条件或事件)”可以被解释成为“当确定时”或“响应于确定”或“当检测(陈述的条件或事件)时”或“响应于检测(陈述的条件或事件)”。
实施例一
本发明实施例给出一种实体推荐方法,请参考图1,其为本发明实施例所提供的实体推荐方法的流程示意图,如图所示,该方法包括以下步骤:
S101,根据获得的查询词,获得至少两个实体集合。
具体的,本发明实施例中,可以接收用户输入的查询词,然后根据该查询词进行搜索,以获得与查询词相匹配的若干搜索结果,最后通过搜索结果页向用户展现若干搜索结果。在生成的搜索结果页中,需要同时向用户提供多维度、多类别的推荐实体。为了实现向用户提供多维度、多类别的推荐实体,本发明实施例中,需要先根据获得的查询词,获得至少两个实体集合。
举例说明,本发明实施例中,根据获得的查询词,获得至少两个实体集合的方法可以包括但不限于以下两种:
第一种:根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合。
可以理解的是,若查询词中只包含搜索实体,或者查询词为搜索实体,说明根据查询词无法确定用户的搜索需求信息,因此用户的搜索需求不明确,则可以从与所述搜索实体相关的所有实体中,获得热度最高
的至少两个实体。
进一步,可以利用聚类算法对热度最高的至少两个实体进行聚类运算,将所述至少两个实体分成至少两个实体集合,从而实现根据获得的查询词,获得至少两个实体集合。
第二种:根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
可以理解的是,若查询词中包含搜索实体和搜索需求信息,说明根据查询词可以确定用户的搜索需求信息,因此用户的搜索需求明确,则可以根据搜索实体和搜索需求信息,获得与搜索需求信息相关度最高的至少两个候选集合。
在一个具体的实现过程中,可以根据各候选集合与搜索需求信息的相关度,并按照相关度由大到小的顺序,对各候选集合进行排序,以获得排序结果,最后从排序结果中获取排序最高的至少两个候选集合,以作为所述至少两个实体集合。
例如,用户输入的查询词为“哈士奇”时,用户的搜索需求不明确,为了激发用户的潜在搜索需求,可以获得与“哈士奇”相关度最高、热度最高的若干实体集合,如“和哈士奇一样智商让人捉急的狗狗”、“和哈士奇一样长得像狼的狗”、“和哈士奇一样性格温顺的无攻击中型犬”。
又例如,用户输入的查询词为“哈士奇好养吗”时,用户的搜索需求很明确,即对哈士奇及其他相关宠物的饲养方式感兴趣,可以获得与用户的搜索需求信息相关度最高的若干实体集合,如“和哈士奇一样较
难饲养的中型犬”、“和哈士奇相反适合家养的小型宠物犬”、“和哈士奇一样忠诚听话好训练的宠物”。
本方法中,将用户输入的查询词的不同搜索需求进行划分,将与搜索需求相匹配的、用户感兴趣的实体按照类别进行有序分组,形成不同的实体集合,最后将这些实体集合推荐给用户,用于满足用户不同的搜索需求,其中,各实体集合的排序可以依据实体集合与用户输入的查询词之间针对搜索需求的相关度,与用户的搜索需求相关度最高的实体集合,排在最靠前的位置。
S102,根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息。
具体的,本发明实施例中,在获得至少两个实体集合后,可以抽取各实体集合与查询词的共同特征,然后根据各实体集合与查询词的共同特征,为各实体集合生成对应的描述信息。
需要说明的是,生成描述信息时,需要在描述信息中构建并体现出与查询词或者搜索意图信息之间存在的关系。例如,查询词为“贝多芬”,实体集合中包含推荐实体,如“霍金”和“文森特·梵高”等,该实体集合与查询词的共同特征为“身残志坚的名人”,则该实体集合对应的描述信息可以为“与贝多芬一样身残志坚的名人”。
另外,实体集合对应的描述信息不仅可以体现出实体集合与搜索词之间的共同特征,还体现出向用户推荐该实体集合的原因,使用户能够更好地理解实体集合中包含的各推荐实体,以及更好地理解推荐的实体集合与用户搜索的查询词之间的关系,或者实体集合与用户的搜索意图之间的关系,从而使用户能够快速选择出感兴趣的或者所需要的实体,
进而提高了用户获取所需要的推荐实体的效率。
S103,在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息。
具体的,本发明实施例中,在为各实体集合生成对应的描述信息后,可以在显示与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息。例如,可以在与所述查询词相匹配的搜索结果页中,任意两条搜索结果的中间位置,显示各实体集合对应的描述信息。
在一个具体的实现过程中,也可以获得与各实体集合相匹配的目标图片;然后,在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
例如,在与查询词相匹配的搜索结果页中,水平显示各实体集合对应的描述信息,以及在显示的每个描述信息的下方位置,显示与该描述信息对应的实体集合相匹配的目标图片。通过各实体集合对应的描述信息与所述目标图片的对应显示,可以将各实体集合对应的描述信息与对应的目标图片关联,使用户通过目标图片可以直接感知到推荐的实体集合。
举例说明,本发明实施例中,获得与各实体集合相匹配的目标图片的方法可以包括但不限于以下两种:
第一种:对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片。
第二种:对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
在一个具体的实现过程中,可以将与实体集合相匹配的多图进行聚合,如可以是三个候选图片、四个候选图片或者五个候选图片的聚合,甚至可以是更多具有代表性的单实体的图片的聚合。利用多图片聚合获得目标图片,可以传递该实体集合中更多推荐实体的图片,还可以增强用户对实体集合的推荐感知。
例如,为实体集合生成多图聚合样式的目标图片时,可以选择该实体集合中,前若干个推荐实体中各推荐实体所对应的图片,然后对这些图片进行排列聚合。或者,又例如,也可以选择实体集合内最能反映该实体集合所代表的搜索需求的图片,并且在图片的视觉、风格、颜色等特征上将相似的图片进行聚合,生成目标图片。
请参考图2(a)~图2(h),其为本发明实施例所提供的至少两个候选图片生成目标图片时的组合方式示例图,如图2(a)~图2(f)所示,至少两个候选图片可以是多个候选图片组合生成目标图片。如图2(g)所示,也可以将至少两个候选图片进行阴影叠加生成目标图片。如图2(h)所示,也可以将至少两个候选图片进行3D组合生成目标图片。需要说明的是,上述利用至少两个候选图片生成目标图片的方式仅为举例说明,本发明实施例对该生成方式不进行特别限定。
S104,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
具体的,本发明实施例中,由于在显示与所述查询词相匹配的搜索结果页中显示了各实体集合对应的描述信息,因此,如果各实体集合中有用户所需要了解的实体集合,则用户可以在各实体集合对应的描述信息中,利用手势向左滑动或者向右滑动实体集合,从而可以浏览到各实
体集合的描述信息。当用户对某个实体集合有兴趣时,用户点击该实体集合的描述信息,这样,可以根据用户所点击的描述信息,进一步显示所点击的描述信息对应的实体集合所包含的各推荐实体,这样,将该实体集合所包含的各推荐实体提供给用户。
举例说明,本发明实施例中,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体的方法可以包括但不限于以下两种:
第一种:利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
在一个具体的实现过程中,可以在所述新页面中利用通栏显示各推荐实体;然后,可以根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
可以理解的是,当用户在各实体集合对应的描述信息中选择一个描述信息后,可以根据所点击的描述信息对应的实体集合,进行一次新的搜索,并在新页面的顶部位置显示一个通栏,该通栏可以固定在搜索结果页的顶部,这样,用户可以在通栏中快速浏览所点击的描述信息对应的实体集合中包含的所有推荐实体。
用户可以进一步点击所需要的推荐实体,根据用户所点击的推荐实体,可以通过异步加载技术,在新页面中通栏下方自动渲染出与所点击的推荐实体相匹配的搜索结果,其中,点击推荐实体、渲染出与所点击的推荐实体相匹配的搜索结果都是在所述新页面中进行,不需要跳转到其他页面,因此无需页面之间的来回切换,从而能够非常方便地点击所
需要的推荐实体,并快速获得与点击的推荐实体相匹配的搜索结果,大大提高了获取多个推荐实体的效率,减少搜索时间。
可选的,还可以在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
可以理解的是,所述通栏可以由两部分组成:导航信息和各推荐实体。导航信息可以显示在通栏中的各推荐实体上方,便于用户快速了解由查询词到点击的推荐实体之间的推荐逻辑,便于用户快速定位所点击的推荐实体。
例如,所述导航信息中所述查询词位于左边,所述推荐实体所在实体集合对应的描述信息位于中间,所点击的推荐实***于右边。例如,导航信息可以为“贝多芬>和贝多芬一样身残志坚的名人>文森特·梵高”,其中,“贝多芬”是用户输入的查询词,“和贝多芬一样身残志坚的名人”是用户在各实体集合中所点击的实体集合对应的描述信息,“文森特·梵高”是用户在该实体集合包含的各推荐信息中进一步点击的一个推荐信息。
另外,也可以对实体集合的描述信息进行信息提炼,将提炼后的精简信息作为导航信息中的描述信息,如导航信息也可以为“贝多芬>类似的身残志坚的名人>文森特·梵高”,从而减少了信息冗余,提高了获取信息的效率。
本方法中,在通栏中显示各推荐实体时,可以水平方向显示各推荐实体的名称以及各推荐实体对应的图片。其中,可以依据各推荐实体与搜索需求信息的相关度,对各推荐实体进行排序,相关度越高的推荐实
体越排在靠前的位置,用户可以在通栏中向左滑动或者向右滑动来浏览各推荐实体。另外,也可以由上述包含各推荐实体的新页面返回与查询词相匹配的搜索结果页,并进一步选择其他的实体集合。
第二种:根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
在一个具体的实现过程中,当用户根据各实体集合对应的描述信息中选择感兴趣的实体集合时,可以利用向左滑动手势或者向右滑动手势来移动各实体集合,继而浏览到各实体集合,当用户点击感兴趣的某实体集合对应的描述信息时,可以在与查询词相匹配的搜索结果页中各实体集合对应的描述信息下方显示下拉框,进而在下拉框中显示用户所点击的描述信息对应的实体集合中所包含的各推荐实体,例如,该实体集合中可以包含12~20个推荐实体,可以根据推荐实体与搜索需求的相关度对各推荐实体进行排序,相关度越高的推荐实体越排在靠前的位置。
进一步的,在下拉框中,用户也可以利用向左滑动手势或者向右滑动手势来移动各推荐实体,以浏览到该实体集合中包含的所有推荐实体,使用户在搜索结果页中能够浏览到各维度的实体集合,以及实体集合中多个推荐实体,提高了获取推荐实体的效率。
实施例二
当用户输入查询词“贝多芬”时(古典主义音乐风格的集大成者,出身贫寒、身残志坚的音乐艺术家),在主搜索需求得到满足的情况下,用户可能感兴趣的其他搜索需求还包括:和贝多芬一样身残志坚的名人、和贝多芬一样出身贫寒的艺术家、和贝多芬拥有一样创作风格的音乐家
等。
请参考图3,其为现有技术中在搜索结果页中推荐实体的第一示例图,如图所示,传统技术方案中,推荐实体围绕贝多芬推荐了“莫扎特”、“肖邦”、“舒伯特”、“巴赫”等若干实体,这些都是世界上知名的音乐家,推荐实体的维度单一且“相关人物”这个描述信息对推荐实体的解释性弱,对于用户在“和贝多芬一样出身贫寒的艺术家”、“和贝多芬一样身残志坚的名人”这两个搜索需求不能满足。
请参考图4(a)~图4(c),其为本发明实施例所提供的在搜索结果页中推荐实体的第一示例图。如图4(a)所示,将“海顿”、“莫扎特”、“肖邦”、“巴赫”等实体归纳在一个实体集合中,提取共同特征并生成描述信息“和贝多芬拥有一样创作风格的音乐家”,与传统技术方案相比,增加了“梵高、霍金、史铁生、阿炳”等组成的实体集合,并为该实体集合增加描述信息“和贝多芬一样身残志坚的名人”,以及增加了“安徒生、徐悲鸿、米开朗琪罗、舒伯特”等组成的实体集合,并为该实体集合增加描述信息“和贝多芬一样出身贫寒的艺术家”,以及,为这三个实体集合分别配上多图聚合后的图片,如图4(a)所示,将各实体集合中最有代表的三个实体的图片进行组合,与传统技术方案相比,本发明实施例所提供的技术方案做到了从实体推荐到实体集合推荐的升级。
如图4(b)所示,当用户对其他和贝多芬拥有一样创作风格的音乐家感兴趣时,点击实体集合的描述信息“和贝多芬拥有一样创作风格的音乐家”,然后会显示新页面,以及在新页面中显示通栏,通栏中显示该实体集合中所包含的所有推荐实体,如海顿、莫扎特、肖邦、巴赫等。
用户通过选择点击感兴趣的推荐实体,当前页面将显示该推荐实体的搜索结果。
当用户想继续浏览其他实体集合时可以返回原搜索结果页,并进一步点击描述信息“和贝多芬一样身残志坚的名人”或者“和贝多芬一样出身贫寒的艺术家”,如图4(c)所示,可以再由原搜索结果页跳转到显示“和贝多芬一样身残志坚的名人”对应的实体集合中所包含的推荐实体的新页面。
请参考图5(a)~图5(b),其为本发明实施例所提供的在搜索结果页中推荐实体的第二示例图。如图5(a)所示,若用户对“和贝多芬拥有一样创作风格的音乐家”感兴趣时,点击该描述信息,在与查询词相关的搜索结果页中显示的各实体模型下方显示下拉框,在下拉框中显示该描述信息对应的实体集合中的各推荐实体,如“莫扎特”、“肖邦”、“海顿”和“巴赫”等推荐实体。另外,如图5(a)所示,用户可以在各推荐实体上利用向左滑动手势或者向右滑动手势来浏览到所有推荐实体。如图5(b)所示,若用户对“和贝多芬一样身残志坚的名人”感兴趣时,点击该描述信息,在与查询词相关的搜索结果页中显示的各实体模型下方显示下拉框,在下拉框中显示该描述信息对应的实体集合中的各推荐实体,如“梵高”、“霍金”、“史铁生”和“阿炳”等推荐实体。
实施例三
当用户输入查询词“阿甘正传”时(汤姆汉克斯主演的励志电影,1995年获奥斯卡6项大奖),在主搜索需求得到满足的情况下,用户可能感兴趣的其他搜索需求还包括:其他获得奥斯卡奖的电影作品、经典
的励志电影、汤姆汉克斯主演的其他作品。
请参考图6,其为现有技术中在搜索结果页中推荐实体的第二示例图,如图所示,传统技术方案中,推荐实体围绕阿甘正装推荐了“肖申克的救赎”、“追梦赤子心”、“成事在人”等若干实体,这些都是获得奥斯卡的著名电影,推荐实体的维度单一且“相关推荐”这个描述信息对推荐实体的解释性弱,对于用户的其他搜索需求不能满足。
请参考图7(a)~图7(c),其为本发明实施例所提供的在搜索结果页中推荐实体的第三示例图。如图7(a)所示,将“鸟人”、“飞越疯人院”、“燃情岁月”、“少年派的奇幻漂流”等实体归纳在一个实体集合中,提取共同特征并生成描述信息“和阿甘正传一样高票房的奥斯卡获奖作品”。将“肖申克的救赎”、“当幸福来敲门”、“永不放弃”、“放牛班的春天”等组成的实体集合,并为该实体集合增加描述信息“和阿甘正传一样经典必看的励志电影”。另外,还可以增加“美人鱼”、“菲利普船长”、“费城故事”和“荒岛余生”等组成的实体集合,并为该实体集合增加描述信息“同为汤姆汉克斯主演的热门电影”。
如图7(a)所示,为这三个实体集合分别配上多图聚合后的图片,将各实体集合中最有代表的五个实体的图片进行组合,与传统技术方案相比,本发明实施例所提供的技术方案做到了从实体推荐到实体集合推荐的升级。
如图7(b)所示,当用户对其他经典励志电影感兴趣时,点击实体集合的描述信息“和阿甘正传一样经典必看的励志电影”,然后会显示新页面,以及在新页面中显示通栏,通栏中显示该实体集合中所包含的所有推荐实体,如“肖申克的救赎”、“当幸福来敲门”、“永不放弃”、
“放牛班的春天”等。用户通过选择点击感兴趣的推荐实体,当前页面将显示该推荐实体的搜索结果。
当用户想继续浏览其他实体集合时可以返回原搜索结果页,并进一步点击描述信息“和阿甘正传一样高票房的奥斯卡获奖作品”,如图7(c)所示,可以再由原搜索结果页跳转到显示“和阿甘正传一样高票房的奥斯卡获奖作品”对应的实体集合中所包含的推荐实体的新页面。
请参考图8(a)~图8(b),其为本发明实施例所提供的在搜索结果页中推荐实体的第四示例图。如图8(a)所示,若用户对“和阿甘正传一样经典必看的励志电影”感兴趣时,点击该描述信息,在与查询词相关的搜索结果页中显示的各实体模型下方显示下拉框,在下拉框中显示该描述信息对应的实体集合中的各推荐实体,如“永不妥协”、“乔丹传人”、“后继有人”和“弱点”等推荐实体。如图8(b)所示,若用户对“和阿甘正传一样高票房的奥斯卡获奖作品”感兴趣时,点击该描述信息,在与查询词相关的搜索结果页中显示的各实体模型下方显示下拉框,在下拉框中显示该描述信息对应的实体集合中的各推荐实体,如“鸟人”、“盗梦空间”、“少年派的奇幻漂流”、“模仿游戏”等推荐实体。
本发明实施例进一步给出实现上述方法实施例中各步骤及方法的装置实施例。
请参考图9,其为本发明实施例所提供的实体推荐装置的功能方块图。如图所示,该装置包括:
实体获取模块91,用于根据获得的查询词,获得至少两个实体集合;
信息获取模块92,用于根据各实体集合与所述查询词的共同特征,
为各实体集合生成对应的描述信息;
推荐显示模块93,用于在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
在一个具体的实现过程中,所述实体获取模块91,具体用于:
根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合;或者,
根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
在一个具体的实现过程中,所述装置还可以包括:图片生成模块94,用于获得与各实体集合相匹配的目标图片;
所述推荐显示模块93,具体用于:在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
在一个具体的实现过程中,所述图片生成模块94,具体用于:
对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片;或者,
对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
在一个具体的实现过程中,所述推荐显示模块93,具体用于:
利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
在一个具体的实现过程中,所述新页面中利用通栏显示各推荐实体;所述推荐显示模块93,还用于:
根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
在一个具体的实现过程中,所述推荐显示模块93,还用于:在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;
其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
或者,在一个具体的实现过程中,所述推荐显示模块93,具体用于:
根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
由于本实施例中的各单元能够执行图1所示的方法,本实施例未详细描述的部分,可参考对图1的相关说明。
本发明实施例的技术方案具有以下有益效果:
本发明实施例中,通过根据获得的查询词,获得至少两个实体集合;从而,根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;进而,在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
本发明实施例提供的技术方案能够在搜索结果页中提供多个实体集
合的描述信息,并且能够进一步显示实体集合所包含的各推荐实体,与现有技术中只能提供类别单一的推荐实体的技术方案相比,本发明实施例所提供的技术方案能够输出多类别、数量多的推荐实体,从而提高了搜索结果页中推荐实体的多样性。而且,与现有技术中推荐实体与查询词之间推荐关系不明确的技术方案相比,本发明实施例中通过输出实体集合对应的描述信息,可以输出查询词与推荐实体之间的推荐关系,帮助用户快速获得所需要的实体,提高了从搜索结果页中获取推荐实体的效率。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的***,装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
在本发明所提供的几个实施例中,应该理解到,所揭露的***,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述单元的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式,例如,多个单元或组件可以结合或者可以集成到另一个***,或一些特征可以忽略,或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本发明各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元,可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中,包括若干指令用以使得一台计算机装置(可以是个人计算机,服务器,或者网络装置等)或处理器(Processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本发明保护的范围之内。
Claims (18)
- 一种实体推荐方法,其特征在于,所述方法包括:根据获得的查询词,获得至少两个实体集合;根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
- 根据权利要求1所述的方法,其特征在于,所述根据获得的查询词,获得至少两个实体集合,包括:根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合;或者,根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
- 根据权利要求1所述的方法,其特征在于,在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,包括:获得与各实体集合相匹配的目标图片;在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
- 根据权利要求3所述的方法,其特征在于,获得与各实体集合 相匹配的目标图片,包括:对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片;或者,对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
- 根据权利要求1所述的方法,其特征在于,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体,包括:利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
- 根据权利要求5所述的方法,其特征在于,所述新页面中利用通栏显示各推荐实体;所述方法还包括:根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
- 根据权利要求6所述的方法,其特征在于,所述方法还包括:在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
- 根据权利要求1所述的方法,其特征在于,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体,包括:根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页 中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
- 一种实体推荐装置,其特征在于,所述装置包括:实体获取模块,用于根据获得的查询词,获得至少两个实体集合;信息获取模块,用于根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;推荐显示模块,用于在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
- 根据权利要求9所述的装置,其特征在于,所述实体获取模块,具体用于:根据所述查询词中包含的搜索实体,在与所述搜索实体相关的实体中获得热度最高的至少两个实体;以及,对所述至少两个实体进行聚类,以获得所述至少两个实体集合;或者,根据所述查询词中包含的搜索实体和搜索需求信息,获得与所述搜索实体和所述搜索需求信息相匹配的至少两个候选集合,根据各候选集合与所述搜索需求信息的相关度,从至少两个候选集合中获得所述至少两个实体集合。
- 根据权利要求9所述的装置,其特征在于,所述装置还包括:图片生成模块,用于获得与各实体集合相匹配的目标图片;所述推荐显示模块,具体用于:在与所述查询词相匹配的搜索结果页中,进行各实体集合对应的描述信息与所述目标图片的对应显示。
- 根据权利要求11所述的装置,其特征在于,所述图片生成模块,具体用于:对于每个所述实体集合,获得与所述实体集合相匹配的一个目标图片;或者,对于每个所述实体集合,获得与所述实体集合相匹配的至少两个候选图片,对所述至少两个候选图片进行拼图,以生成与所述实体集合相匹配的目标图片。
- 根据权利要求9所述的装置,其特征在于,所述推荐显示模块,具体用于:利用用户所点击的描述信息对应的实体集合所包含的各推荐实体进行页面渲染,以展现包含各推荐实体的新页面。
- 根据权利要求13所述的装置,其特征在于,所述新页面中利用通栏显示各推荐实体;所述推荐显示模块,还用于:根据用户在所述通栏中显示的各推荐实体中所点击的推荐实体,在所述新页面中所述通栏下方显示与所点击的推荐实体相匹配的搜索结果。
- 根据权利要求14所述的装置,其特征在于,所述推荐显示模块,还用于:在所述通栏中显示所点击的推荐实体与所述查询词之间的导航信息;其中,所述导航信息包括所述查询词、所述推荐实体所在实体集合对应的描述信息以及所点击的推荐实体。
- 根据权利要求9所述的装置,其特征在于,所述推荐显示模块,具体用于:根据用户所点击的描述信息,在与所述查询词相匹配的搜索结果页 中所述描述信息下方显示下拉框,以及,利用所述下拉框显示用户所点击的描述信息对应的实体集合所包含的各推荐实体。
- 一种设备,包括一个或者多个处理器;存储器;一个或者多个程序,所述一个或者多个程序存储在所述存储器中,当被所述一个或者多个处理器执行时:根据获得的查询词,获得至少两个实体集合;根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
- 一种计算机存储介质,所述计算机存储介质被编码有计算机程序,所述程序在被一个或多个计算机执行时,使得所述一个或多个计算机执行如下操作:根据获得的查询词,获得至少两个实体集合;根据各实体集合与所述查询词的共同特征,为各实体集合生成对应的描述信息;在与所述查询词相匹配的搜索结果页中显示各实体集合对应的描述信息,以及,根据用户所点击的描述信息,显示所点击的描述信息对应的实体集合所包含的各推荐实体。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510828555.X | 2015-11-25 | ||
CN201510828555.XA CN105512178B (zh) | 2015-11-25 | 2015-11-25 | 一种实体推荐方法及装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2017088497A1 true WO2017088497A1 (zh) | 2017-06-01 |
Family
ID=55720160
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2016/089649 WO2017088497A1 (zh) | 2015-11-25 | 2016-07-11 | 一种实体推荐方法、装置、设备及计算机存储介质 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN105512178B (zh) |
WO (1) | WO2017088497A1 (zh) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110929139A (zh) * | 2018-09-04 | 2020-03-27 | 阿里巴巴集团控股有限公司 | 搜索推荐方法及装置 |
CN111680230A (zh) * | 2020-05-25 | 2020-09-18 | 北京百度网讯科技有限公司 | 搜索结果页的展示方法、装置、电子设备及存储介质 |
CN111797257A (zh) * | 2020-06-29 | 2020-10-20 | 京东方科技集团股份有限公司 | 基于词向量的图片推荐方法及相关设备 |
CN114064930A (zh) * | 2021-11-29 | 2022-02-18 | 北京字节跳动网络技术有限公司 | 一种数据展示方法、装置、计算机设备以及存储介质 |
CN114722823A (zh) * | 2022-03-24 | 2022-07-08 | 华中科技大学 | 构建航空知识图谱的方法及装置、计算机可读介质 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105512178B (zh) * | 2015-11-25 | 2019-02-05 | 百度在线网络技术(北京)有限公司 | 一种实体推荐方法及装置 |
CN111949864B (zh) * | 2020-08-10 | 2022-02-25 | 北京字节跳动网络技术有限公司 | 一种搜索方法、装置、电子设备及存储介质 |
CN116483946A (zh) * | 2022-01-14 | 2023-07-25 | 腾讯科技(深圳)有限公司 | 数据处理方法、装置、设备及计算机程序产品 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101364239A (zh) * | 2008-10-13 | 2009-02-11 | 中国科学院计算技术研究所 | 一种分类目录自动构建方法及相关*** |
CN102609433A (zh) * | 2011-12-16 | 2012-07-25 | 北京大学 | 基于用户日志进行查询推荐的方法及*** |
CN104281699A (zh) * | 2014-10-15 | 2015-01-14 | 百度在线网络技术(北京)有限公司 | 搜索推荐方法及装置 |
CN104778255A (zh) * | 2015-04-20 | 2015-07-15 | 百度在线网络技术(北京)有限公司 | 搜索结果的推荐方法和装置 |
US20150234813A1 (en) * | 2013-11-04 | 2015-08-20 | Michael R. Knapp | Systems and Methods for Categorizing and Accessing Information Databases and for Displaying Query Results |
CN105512178A (zh) * | 2015-11-25 | 2016-04-20 | 百度在线网络技术(北京)有限公司 | 一种实体推荐方法及装置 |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6834308B1 (en) * | 2000-02-17 | 2004-12-21 | Audible Magic Corporation | Method and apparatus for identifying media content presented on a media playing device |
US20110307460A1 (en) * | 2010-06-09 | 2011-12-15 | Microsoft Corporation | Navigating relationships among entities |
US9092527B2 (en) * | 2013-01-30 | 2015-07-28 | Quixey, Inc. | Performing application search based on entities |
-
2015
- 2015-11-25 CN CN201510828555.XA patent/CN105512178B/zh active Active
-
2016
- 2016-07-11 WO PCT/CN2016/089649 patent/WO2017088497A1/zh active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101364239A (zh) * | 2008-10-13 | 2009-02-11 | 中国科学院计算技术研究所 | 一种分类目录自动构建方法及相关*** |
CN102609433A (zh) * | 2011-12-16 | 2012-07-25 | 北京大学 | 基于用户日志进行查询推荐的方法及*** |
US20150234813A1 (en) * | 2013-11-04 | 2015-08-20 | Michael R. Knapp | Systems and Methods for Categorizing and Accessing Information Databases and for Displaying Query Results |
CN104281699A (zh) * | 2014-10-15 | 2015-01-14 | 百度在线网络技术(北京)有限公司 | 搜索推荐方法及装置 |
CN104778255A (zh) * | 2015-04-20 | 2015-07-15 | 百度在线网络技术(北京)有限公司 | 搜索结果的推荐方法和装置 |
CN105512178A (zh) * | 2015-11-25 | 2016-04-20 | 百度在线网络技术(北京)有限公司 | 一种实体推荐方法及装置 |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110929139A (zh) * | 2018-09-04 | 2020-03-27 | 阿里巴巴集团控股有限公司 | 搜索推荐方法及装置 |
CN110929139B (zh) * | 2018-09-04 | 2023-12-29 | 阿里巴巴集团控股有限公司 | 搜索推荐方法及装置 |
CN111680230A (zh) * | 2020-05-25 | 2020-09-18 | 北京百度网讯科技有限公司 | 搜索结果页的展示方法、装置、电子设备及存储介质 |
CN111680230B (zh) * | 2020-05-25 | 2023-04-18 | 北京百度网讯科技有限公司 | 搜索结果页的展示方法、装置、电子设备及存储介质 |
CN111797257A (zh) * | 2020-06-29 | 2020-10-20 | 京东方科技集团股份有限公司 | 基于词向量的图片推荐方法及相关设备 |
CN111797257B (zh) * | 2020-06-29 | 2024-02-23 | 京东方科技集团股份有限公司 | 基于词向量的图片推荐方法及相关设备 |
CN114064930A (zh) * | 2021-11-29 | 2022-02-18 | 北京字节跳动网络技术有限公司 | 一种数据展示方法、装置、计算机设备以及存储介质 |
CN114722823A (zh) * | 2022-03-24 | 2022-07-08 | 华中科技大学 | 构建航空知识图谱的方法及装置、计算机可读介质 |
CN114722823B (zh) * | 2022-03-24 | 2023-04-14 | 华中科技大学 | 构建航空知识图谱的方法及装置、计算机可读介质 |
Also Published As
Publication number | Publication date |
---|---|
CN105512178B (zh) | 2019-02-05 |
CN105512178A (zh) | 2016-04-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017088497A1 (zh) | 一种实体推荐方法、装置、设备及计算机存储介质 | |
Zheng et al. | Content-aware generative modeling of graphic design layouts | |
Lin et al. | Explainable outfit recommendation with joint outfit matching and comment generation | |
WO2022048360A1 (zh) | 一种搜索结果展示的方法、装置及计算机存储介质 | |
US20230192947A1 (en) | System and method allowing social fashion selection in an electronic marketplace | |
CA2849293C (en) | Providing topic based search guidance | |
US9607010B1 (en) | Techniques for shape-based search of content | |
US8725717B2 (en) | System and method for identifying topics for short text communications | |
WO2016091044A1 (zh) | 一种热词推荐方法、装置、***、设备和计算机存储介质 | |
US20180181569A1 (en) | Visual category representation with diverse ranking | |
US9342601B1 (en) | Query formulation and search in the context of a displayed document | |
WO2019113977A1 (zh) | 文章处理方法、装置、服务器及存储介质 | |
US9727906B1 (en) | Generating item clusters based on aggregated search history data | |
US20140280289A1 (en) | Autosuggestions based on user history | |
US20110191336A1 (en) | Contextual image search | |
TW201510753A (zh) | 查詢建議模板 | |
US9324102B2 (en) | System and method to retrieve relevant inventory using sketch-based query | |
Choi et al. | Fontmatcher: font image paring for harmonious digital graphic design | |
CN113330455A (zh) | 使用有条件的生成对抗网络查找互补的数字图像 | |
WO2016107125A1 (zh) | 信息搜索方法及装置 | |
EP4231172A1 (en) | Aspect-aware autocomplete query | |
Wang et al. | JIGSAW: interactive mobile visual search with multimodal queries | |
US11282124B1 (en) | Automated identification of item attributes relevant to a browsing session | |
CN106951433B (zh) | 一种检索方法及装置 | |
Lu et al. | Browse-to-search: Interactive exploratory search with visual entities |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16867724 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 16867724 Country of ref document: EP Kind code of ref document: A1 |