Article recommendation method and device

Abstract

The invention discloses an article recommendation method and device, and relates to the technical field of internet medical treatment. A specific embodiment of the method comprises the following steps: extracting key information from inquiry dialogue data of a user; acquiring an article matched with the key information from historical interaction behavior data of the user as a reference article; determining a recommended article set corresponding to the user according to the article similarity between each other article and the reference article; the other articles are articles which are matched with the key information and are different from the reference article. According to the embodiment, the recommended item corresponding to the user can be automatically determined by analyzing the inquiry dialogue data, the management efficiency and timeliness are improved, and personalized recommendation is achieved.

Description

Technical Field

[0001] The present application relates to the field of Internet medical technology, and in particular to a method and device for item recommendation. Background Art

[0002] With the promotion and popularization of smart medical platforms, doctors and patients can communicate quickly through the Internet platform and realize online consultation. Compared with offline consultation, online consultation has the advantages of breaking through geographical restrictions, being fast and efficient, and convenient post-diagnosis management. In online consultation, post-diagnosis management is mainly carried out through one-on-one diet and medication guidance from doctors to patients to manage the course of disease, which has the problems of low management efficiency and poor timeliness. Summary of the invention

[0003] To this end, the present invention provides a method and device for item recommendation, which can automatically determine the recommended items corresponding to the user by analyzing the consultation dialogue data, improve management efficiency and timeliness, and realize personalized recommendations.

[0004] According to a first aspect of an embodiment of the present invention, a method for item recommendation is provided, comprising: extracting key information from a user's medical consultation dialogue data; acquiring items matching the key information from the user's historical interaction behavior data as reference items; determining a set of recommended items corresponding to the user based on the item similarity between each other item and the reference item; the other items are items matching the key information and different from the reference items.

[0005] Optionally, extracting key information from the user's medical consultation dialogue data includes: performing word segmentation processing on the medical consultation dialogue data, classifying words in the segmented medical consultation dialogue data using a preset sequence labeling model, and extracting the key information from the classification results.

[0006] Optionally, the key information includes at least one of the following: disease name, symptoms, and medication.

[0007] Optionally, obtaining items matching the key information from the historical behavior records of the user as reference items includes: obtaining a set of candidate items matching the key information from the historical interaction behavior data of the user, and screening candidate items whose scores meet preset score recommendations from the candidate item set as the reference items.

[0008] Optionally, determining a recommended item set corresponding to the user according to the item similarity between each other item and the reference item includes: taking other items that meet at least one of the following conditions as recommended items: the item similarity meets a preset similarity condition, the comprehensive score meets a preset comprehensive score condition, and the items have item attributes that the user is concerned about;

[0009] The comprehensive score of each of the other items is determined based on a pre-constructed user-item score matrix, which contains scores of different users on each of the other items; the item attributes that the user is concerned about are determined based on the evaluation records of the reference items obtained from the user's historical interaction behavior data.

[0010] Optionally, the method further includes: after determining the recommended item set corresponding to the user based on the item similarity between each other item and the reference item, determining the content similarity between the user feature vector of the user and each recommended item in the recommended item set, sorting the items in the recommended item set according to the content similarity, and generating an item recommendation list.

[0011] Optionally, the method further includes: embedding the item recommendation into the user's consultation dialogue page.

[0012] According to a second aspect of an embodiment of the present invention, there is provided an apparatus for recommending items, comprising:

[0013] Key information extraction module, extracting key information from the user's consultation dialogue data;

[0014] A reference item extraction module, which obtains items matching the key information from the historical interaction behavior data of the user as reference items;

[0015] The recommended item determination module determines a recommended item set corresponding to the user according to the item similarity between each other item and the reference item; the other items are items that match the key information and are different from the reference items.

[0016] Optionally, the key information extraction module extracts key information from the user's medical consultation dialogue data, including: performing word segmentation on the medical consultation dialogue data, classifying words in the segmented medical consultation dialogue data using a preset sequence labeling model, and extracting the key information from the classification results.

[0017] Optionally, the key information includes at least one of the following: disease name, symptoms, and medication.

[0018] Optionally, the reference item extraction module obtains items matching the key information from the user's historical behavior records as reference items, including: obtaining a set of candidate items matching the key information from the user's historical interaction behavior data, and screening candidate items whose scores meet preset score recommendations from the candidate item set as the reference items.

[0019] Optionally, the recommended item determination module determines a recommended item set corresponding to the user according to item similarities between each other item and the reference item, including:

[0020] Other items that meet at least one of the following conditions are recommended as items: the item similarity meets the preset similarity condition, the comprehensive score meets the preset comprehensive score condition, and the item has the item attributes that the user is concerned about;

[0021] The comprehensive score of each of the other items is determined based on a pre-constructed user-item score matrix, which contains scores of different users on each of the other items; the item attributes that the user is concerned about are determined based on the evaluation records of the reference items obtained from the user's historical interaction behavior data.

[0022] Optionally, the recommended item determination module is further configured to:

[0023] After determining a recommended item set corresponding to the user based on the item similarities between each other item and the reference item, determine the content similarity between the user feature vector of the user and each recommended item in the recommended item set, sort the items in the recommended item set based on the content similarities, and generate an item recommendation list.

[0024] Optionally, the device further comprises a recommended item embedding module, which is used to embed the item recommendation into the user's consultation dialogue page.

[0025] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for item recommendation, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the first aspect of the embodiment of the present invention.

[0026] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0027] An embodiment of the above invention has the following advantages or beneficial effects: the embodiment of the present invention extracts key information from the user's medical consultation conversation data, obtains items matching the key information from the user's historical interaction behavior data as reference items, and determines the recommended item set corresponding to the user based on the item similarity between each other item and the reference item. The recommended items corresponding to the user can be automatically determined by analyzing the medical consultation conversation data. On the one hand, it can free up manpower and improve management efficiency and timeliness. On the other hand, it can also provide users with more personalized professional services and improve user consultation results and satisfaction.

[0028] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0030] Figure 1 is a schematic diagram of the main process of the method for item recommendation according to an embodiment of the present invention;

[0031] Figure 2 is a schematic diagram of main modules of an apparatus for item recommendation according to an embodiment of the present invention;

[0032] Figure 3 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0033] Figure 4 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0035] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.

[0036] According to a first aspect of an embodiment of the present invention, a method for recommending an item is provided. Figure 1 FIG. 1 is a schematic diagram of the main process of the method for recommending items according to an embodiment of the present invention. Figure 1 As shown, the method for item recommendation includes step S101, step S102 and step S103.

[0037] Step S101: extract key information from the user's medical consultation dialogue data.

[0038] The consultation dialogue data refers to the online conversation data between the user and the doctor during the online consultation. Key information refers to the information related to the core demands of the user in the conversation, and its specific content can be set selectively. Optionally, the key information includes at least one of the following: disease name, symptoms, and medication.

[0039] The implementation method of extracting key information from the user's medical consultation dialogue data can be set according to actual conditions. In some optional embodiments, a key information data set can be pre-set. When extracting key information from the medical consultation dialogue data, the medical consultation dialogue data is first segmented. If there is a word that hits the key information data set in the word segmentation processing, the word is used as the key information. In other optional embodiments, a large number of labeled samples can be used to train a sequence annotation model that can identify key information, and the trained sequence annotation model can be used to identify key information in the medical consultation dialogue data. In some other optional embodiments, extracting key information from the user's medical consultation dialogue data includes: first segmenting the medical consultation dialogue data, and then using a preset sequence annotation model to classify the words in the medical consultation dialogue data after segmentation, and extracting the key information from the classification results.

[0040] In actual application, a model for extracting key information can be built through the following steps:

[0041] 1. Data collection and processing: Establish a database to store information of different types of items, collect the online consultation dialogue data of multiple users to construct a doctor-patient dialogue data set, and perform text cleaning and text segmentation on the consultation dialogue data set. The items in the embodiment of the present invention can be physical items, such as cough medicine and other medicines, or honey, wolfberry and other food supplements; the items in the embodiment of the present invention can also be virtual items, such as online consultation products and other services. Through text cleaning, some irrelevant information that may exist in the data is filtered out, such as HTML (Hypertext Markup Language) tags, URL (Uniform Resource Locator) links, modal particles, etc. Through text segmentation, a continuous text is divided into individual words or phrases, and open source libraries such as jieba, nltk, spaCy, etc. can be used to implement word segmentation.

[0042] 2. Text information extraction: Since the consultation dialogue data is mainly in text form, NLP (Natural Language Processing) technology is mainly used to extract key information.

[0043] First, a data conversion model is built based on the constructed consultation dialogue dataset. The model can be built based on BERT (Bidirectional Encoder Representations from Transformers) or XLNet (Generalized Autoregressive Pretraining for Language Understanding), which is used to segment the sample data and convert each word into a vector representation. BERT is a bidirectional Transformer encoder that can generate a global word representation based on the context of all words. It can learn language knowledge in a large-scale unsupervised corpus through pre-training and extract word, sentence and paragraph level representations. XLNet is an autoregressive pre-training model that can capture long-distance dependencies by learning to predict a word while considering its context. Pre-training BERT or XLNet takes a long time and requires large computing resources. During training, the doctor-patient dialogue information input by the user is first segmented, and then the word vector embedding layer of BERT or XLNet is used to process it through the Transformer network structure to obtain the vector representation of each word, which contains the information of the word and its context.

[0044] Next, you can use a custom model or a neural network model such as LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit) to capture sequence information, thereby extracting key information from the current user's input. For example, you can train a sequence annotation model that can classify words in a sentence into disease names, symptoms, medications, etc., so that you can extract key information.

[0045] In a specific embodiment, the text content of the doctor-patient dialogue data is as follows: From the following dialogue data, key information can be identified, such as coughing with phlegm, being severe in the morning, etc., to determine that the patient may suffer from chronic bronchitis.

[0046] Doctor: Hello, do you have any discomfort that you would like to consult?

[0047] User: My throat has been dry lately and I have been coughing for a while.

[0048] Doctor: Is the cough dry or does it produce phlegm?

[0049] User: It's a productive cough, and it's usually worse in the morning.

[0050] Doctor: According to your description of symptoms, this may be caused by chronic bronchitis. I suggest you use some herbal cough syrups and oral liquids for clearing heat and detoxification, and drink more honey water to relieve symptoms. In addition, be careful to avoid smoking and contact with irritating gases.

[0051] User: Ok, thank you doctor. Where can I buy these medicines?

[0052] Doctor: You can try to buy it from an online pharmacy. I have provided you with two links. You can click to view and purchase the appropriate medicine: Link 1 (drug link 1), Link 2 (drug link 2). At the same time, please use the medicine according to the instructions and pay attention to the adverse reactions and contraindications of the medicine during use.

[0053] User: Thank you very much for the doctor's advice and recommended drug link. I will buy it as soon as possible and use it according to the instructions.

[0054] Doctor: You're welcome. I wish you a speedy recovery. If you have any other questions, feel free to ask me.

[0055] In order to facilitate subsequent analysis and processing, the embodiment of the present invention can encode the key information after extracting the key information. For example, the text of the consultation dialogue and the description of the item can be converted into a numerical vector with the help of word2vec, Doc2vec, etc. Word2vec is a model for generating word vectors, which can capture the semantic relationship between words. There are two main forms of word2vec models: CBOW (Continuous Bag of Words) and Skip-gram. CBOW uses the context of a word to predict the word, while Skip-gram uses a word to predict its context. Doc2Vec is an extension of Word2Vec. In addition to mapping words to vector space, it can also map documents (such as sentences and paragraphs) to vector space. Its training text is based on the window size), min_count (ignore words with too small frequency), etc., to perform model training. Specifically, the text of multiple historical consultation dialogues can be segmented using libraries such as jieba, and the results after segmentation can be input using the Doc2Vec method in the gensim (a natural language processing tool) library. The empirical parameters are set, and the model is trained. The parameters are adjusted to a certain optimal value according to the model training effect. The text of the conversation between the user and the doctor is input into the trained model. The model can convert the conversation text and item description from text information into a numerical vector form that can be processed by a computer, thus realizing information encoding.

[0056] Step S102: Acquire items matching the key information from the historical interaction behavior data of the user as reference items.

[0057] The match in "items matching the key information" can be evaluated from multiple aspects. When the key information is information such as the name of the disease, symptoms, etc., the items matching the key information may be items that have a therapeutic or alleviating effect on the above-mentioned diseases or symptoms, such as medicines or foods, or online diagnosis and treatment products that provide consulting services related to the above-mentioned diseases or symptoms. When the key information is a medicine, the item matching the key information may be the medicine, or an item with the same or similar efficacy as the medicine. The above is only an exemplary description of the key information and the items matching it. Those skilled in the art may also set other key information and set a specific method for determining the items matching it. In addition, it should be noted that when there are multiple items matching the key information, one can select one from them as a reference item, such as selecting the item corresponding to the most recent interactive behavior as a reference item, or selecting the item with the highest sales volume as a reference item, or selecting the item with the highest user rating as a reference item.

[0058] In an optional embodiment of the present invention, obtaining items matching the key information from the user's historical behavior records as reference items includes: obtaining a candidate item set matching the key information from the user's historical interaction behavior data, and screening candidate items whose scores meet preset score recommendations from the candidate item set as the reference items. The interaction behavior may be any one or more of browsing, adding to a shopping cart, sharing, collecting, placing an order, purchasing, etc.

[0059] Step S103: determining a recommended item set corresponding to the user according to the item similarity between each other item and the reference item; the other items are items that match the key information and are different from the reference items.

[0060] Optionally, determining a recommended item set corresponding to the user according to the item similarity between each other item and the reference item includes: taking other items that meet at least one of the following conditions as recommended items:

[0061] (1) The similarity of the items meets the preset similarity conditions. Exemplarily, the features used to characterize the items are first selected. These features may include the intrinsic attributes of the items, such as product description, category, price, etc., and may also include user attributes, such as age, gender, interests and hobbies, etc. Exemplarily, for honey, its product attributes include brand, category, price, etc. Then the item features are encoded and converted into numerical form, for example, text-type features are converted into numerical vectors through one-hot encoding, TF-IDF, etc. Finally, the similarity between the item feature vectors is calculated, such as cosine similarity, Jaccard coefficient, etc., to obtain the similarity between other items, and other items that meet the preset similarity conditions (for example, greater than or equal to the preset similarity threshold) are recommended items.

[0062] (2) The comprehensive score meets the preset comprehensive score conditions. The comprehensive score of each of the other items is determined based on a pre-constructed user-item score matrix, which contains the scores of different users on each of the other items. A user-item score matrix can be created through the interaction behaviors of multiple users with items (such as purchase records and rating records). The score matrix is ​​transposed so that the rows represent items and the columns represent users. In this way, each row represents the user's score for the item. For each item, the scores of each user on the item are weighted to obtain the comprehensive score of the item. Other items whose comprehensive scores meet the preset comprehensive score conditions (such as greater than or equal to the preset score threshold) are recommended items.

[0063] (3) Item attributes that users are interested in. The item attributes that users are interested in are determined based on the evaluation records of the reference items obtained from the historical interaction behavior data of the user. For example, if the user evaluation tends to favor favorable prices, other items that also have the attribute of affordable prices will be recommended.

[0064] In actual application, the recommended item set can be determined based on any one of the above conditions, or any two or more of the above conditions. Of course, some or all of the above conditions can also be combined with other conditions. For example, for each user, find the items with high historical scores (highest, or higher than the set value) of the user as the reference items of the user, and then find other items in the user-item rating matrix that are most similar to the reference items and have the same user-concerned item attributes as the reference items, and are highly rated as recommended items. If there are multiple items with high historical scores, a weighted score can also be calculated based on the similarity and score of each item, and the reference item or recommended item can be determined based on the weighted score. If there is no score, this item is not used as a scoring criterion, and the optimal solution is recommended based on other features.

[0065] After determining the recommended item set corresponding to the user, the items in the recommended item set can be sorted. For example, the items can be sorted according to sales volume. Optionally, after determining the recommended item set corresponding to the user, the content similarity between the user feature vector of the user and each recommended item in the recommended item set is determined, and the items in the recommended item set are sorted according to the content similarity to generate an item recommendation list. In an embodiment of the present invention, the features of the user and the item can be selected first. These features can include the intrinsic attributes of the item, such as product description, category, price, etc., and can also include the attributes of the user, such as age, gender, hobbies, etc. The features are then encoded and converted into numerical form, for example, text-type features are converted into numerical vectors by one-hot encoding, TF-IDF, etc. The similarity is then calculated: for each user and the recommended item set determined above, the respective feature vectors are first calculated, and then the similarity between the user feature vector and the item feature vector is calculated, such as cosine similarity, to obtain a similarity matrix between the user and each recommended item. Finally, a recommendation list is generated: for each user, the recommended items are sorted from high to low according to their similarity with each recommended item in the collection to generate a recommendation list.

[0066] Optionally, the method further includes: embedding the item recommendation into the user's consultation dialogue page. Exemplarily, during an online consultation, the doctor prescribes a prescription containing cough syrup and heat-clearing and detoxifying oral liquid to the user. Based on information interpretation, the embodiment of the present invention automatically recommends honey as a food supplement product and a post-diagnosis follow-up consultation product for chronic bronchitis, which is convenient for users to conduct long-term observation of this chronic disease after purchase. By embedding the item recommendation into the user's consultation dialogue page for front-end display and interaction, the embodiment of the present invention can provide users with purchase decision support and user feedback mechanism.

[0067] The embodiment of the present invention extracts key information from the user's consultation dialogue data to recommend items. For example, by analyzing the patient's condition, symptoms and diagnosis results, it can provide targeted post-diagnosis goods and consultation product recommendations according to the personalized needs of each user, thereby achieving personalized recommendations. This personalized recommendation can improve the user experience and help users find products that suit them more quickly.

[0068] In an embodiment of the present invention, after a user determines a recommended item, the item recommendation is embedded in the user's medical consultation dialogue page. The user can obtain detailed information of the recommended item through the dialogue page, realize intelligent interpretation and suggestions, and help the user better understand the condition and treatment plan, thereby improving treatment compliance.

[0069] According to a second aspect of an embodiment of the present invention, a device for implementing the above method is provided.

[0070] Figure 2 is a schematic diagram of main modules of an apparatus for recommending items according to an embodiment of the present invention. Figure 2 As shown, the device 200 for recommending items includes:

[0071] Key information extraction module 201, extracting key information from the user's consultation dialogue data;

[0072] A reference item extraction module 202 is configured to obtain items matching the key information from the historical interaction behavior data of the user as reference items;

[0073] The recommended item determination module 203 determines a recommended item set corresponding to the user according to the item similarity between each other item and the reference item; the other items are items that match the key information and are different from the reference items.

[0074] Optionally, the key information extraction module extracts key information from the user's medical consultation dialogue data, including: performing word segmentation on the medical consultation dialogue data, classifying words in the segmented medical consultation dialogue data using a preset sequence labeling model, and extracting the key information from the classification results.

[0075] Optionally, the key information includes at least one of the following: disease name, symptoms, and medication.

[0076] Optionally, the reference item extraction module obtains items matching the key information from the user's historical behavior records as reference items, including: obtaining a set of candidate items matching the key information from the user's historical interaction behavior data, and screening candidate items whose scores meet preset score recommendations from the candidate item set as the reference items.

[0077] Optionally, the recommended item determination module determines a recommended item set corresponding to the user according to item similarities between each other item and the reference item, including:

[0078] Other items that meet at least one of the following conditions are recommended as items: the item similarity meets the preset similarity condition, the comprehensive score meets the preset comprehensive score condition, and the item has the item attributes that the user is concerned about;

[0079] The comprehensive score of each of the other items is determined based on a pre-constructed user-item score matrix, which contains scores of different users on each of the other items; the item attributes that the user is concerned about are determined based on the evaluation records of the reference items obtained from the user's historical interaction behavior data.

[0080] Optionally, the recommended item determination module is further configured to:

[0081] After determining a recommended item set corresponding to the user based on the item similarities between each other item and the reference item, determine the content similarity between the user feature vector of the user and each recommended item in the recommended item set, sort the items in the recommended item set based on the content similarities, and generate an item recommendation list.

[0082] Optionally, the device further comprises a recommended item embedding module, which is used to embed the item recommendation into the user's consultation dialogue page.

[0083] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for item recommendation, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the first aspect of the embodiment of the present invention.

[0084] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0085] Figure 3 An exemplary system architecture 300 is shown to which the method for item recommendation or the apparatus for item recommendation according to an embodiment of the present invention may be applied.

[0086] like Figure 3 As shown, the system architecture 300 may include terminal devices 301, 302, 303, a network 304 and a server 305. The network 304 is used to provide a medium for communication links between the terminal devices 301, 302, 303 and the server 305. The network 304 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0087] Users can use terminal devices 301, 302, 303 to interact with server 305 through network 304 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 301, 302, 303, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).

[0088] The terminal devices 301 , 302 , and 303 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0089] The server 305 may be a server that provides various services, such as a backend management server (only an example) that provides support for shopping websites browsed by users using the terminal devices 301, 302, and 303. The backend management server may analyze and process the received product information query request and other data, and feed back the processing results (such as target push information, product information - only an example) to the terminal device.

[0090] It should be noted that the method for recommending items provided in the embodiment of the present invention is generally executed by the server 305 , and accordingly, the device for recommending items is generally disposed in the server 305 .

[0091] It should be understood that Figure 3 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0092] Reference below Figure 4 , which shows a schematic diagram of the structure of a computer system 400 of a terminal device suitable for implementing an embodiment of the present invention. Figure 4 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0093] like Figure 4 As shown, the computer system 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage part 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the system 400 are also stored. The CPU 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0094] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed, so that a computer program read therefrom is installed into the storage section 408 as needed.

[0095] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 409, and / or installed from the removable medium 411. When the computer program is executed by the central processing unit (CPU) 401, the above-mentioned functions defined in the system of the present invention are executed.

[0096] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0097] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0098] The modules involved in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor including: a key information extraction module, a reference item extraction module and a recommended item determination module. The names of these modules do not constitute a limitation on the modules themselves in some cases. For example, the key information extraction module may also be described as "a module for determining a recommended item set corresponding to the user based on the item similarity between each other item and the reference item".

[0099] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: extracting key information from the user's medical consultation dialogue data; obtaining items matching the key information from the user's historical interaction behavior data as reference items; determining a recommended item set corresponding to the user based on the item similarity between each other item and the reference item; the other items are items matching the key information and different from the reference items.

[0100] According to the technical solution of the embodiment of the present invention, by extracting key information from the user's medical consultation dialogue data, obtaining items matching the key information from the user's historical interaction behavior data as reference items, and determining the recommended item set corresponding to the user based on the item similarity between each other item and the reference item, the recommended items corresponding to the user can be automatically determined by analyzing the medical consultation dialogue data. On the one hand, it can free up manpower and improve management efficiency and timeliness. On the other hand, it can also provide users with more personalized professional services and improve user consultation results and satisfaction.

[0101] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for item recommendation, characterized in that: include: Extract key information from the user's consultation conversation data; Acquire items matching the key information from the historical interaction behavior data of the user as reference items; A recommended item set corresponding to the user is determined according to the item similarity between each other item and the reference item; the other items are items that match the key information and are different from the reference items.

2. The method according to claim 1, characterized in that Extract key information from the user's consultation conversation data, including: The medical consultation dialogue data is segmented, the words in the segmented medical consultation dialogue data are classified using a preset sequence labeling model, and the key information is extracted from the classification results.

3. The method according to claim 1, characterized in that The key information includes at least one of the following: disease name, symptoms, and medication.

4. The method according to claim 1, characterized in that Acquiring an item matching the key information from the historical behavior record of the user as a reference item includes: A candidate item set matching the key information is obtained from the historical interaction behavior data of the user, and candidate items with scores satisfying preset score recommendations are screened from the candidate item set as the reference items.

5. The method according to claim 1, characterized in that Determining a recommended item set corresponding to the user according to the item similarity between each other item and the reference item includes: Other items that meet at least one of the following conditions are recommended as items: the item similarity meets the preset similarity condition, the comprehensive score meets the preset comprehensive score condition, and the item has the item attributes that the user is concerned about; The comprehensive score of each of the other items is determined based on a pre-constructed user-item score matrix, which contains scores of different users on each of the other items; the item attributes that the user is concerned about are determined based on the evaluation records of the reference items obtained from the user's historical interaction behavior data.

6. The method according to claim 1, characterized in that The method further comprises: After determining a recommended item set corresponding to the user based on the item similarities between each other item and the reference item, determine the content similarity between the user feature vector of the user and each recommended item in the recommended item set, sort the items in the recommended item set based on the content similarities, and generate an item recommendation list.

7. The method according to any one of claims 1 to 6, characterized in that: The method also includes: embedding the item recommendation into the user's consultation dialogue page.

8. A device for recommending items, characterized in that: include: Key information extraction module, extracting key information from the user's consultation dialogue data; A reference item extraction module, which obtains items matching the key information from the historical interaction behavior data of the user as reference items; The recommended item determination module determines a recommended item set corresponding to the user according to the item similarity between each other item and the reference item; the other items are items that match the key information and are different from the reference items.

9. An electronic device for recommending items, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images