Live broadcast control method, apparatus and device, and storage medium

Abstract

The invention provides a live broadcast control method, device and equipment and a storage medium, relates to the technical field of computers, in particular to the fields of data processing, artificial intelligence, deep learning, intelligent e-commerce and the like, and can be applied to application scenes such as intelligent assistants, virtual assistants, intelligent e-commerce and the like. The method comprises the following steps: acquiring user interaction information of a live broadcasting room; performing intention recognition on the user interaction information to obtain a user intention; the user intention comprises an identifier of a first commodity required to be explained by the user and a concern point of the user on the first commodity; obtaining commodity details of a first commodity based on the user intention; generating a live script of the first commodity according to the commodity details and the concern; and modifying the live broadcast room script by using the live broadcast script of the first commodity, and playing the modified live broadcast room script. The experience effect of the user in the live broadcast room can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to the fields of data processing, artificial intelligence, deep learning, smart e-commerce, etc., and can be applied to application scenarios such as smart assistants, virtual assistants, and smart e-commerce. Background Art

[0002] In recent years, with the rapid development of artificial intelligence technology, the application of digital human technology in commodity live broadcast business scenarios has become increasingly mature. However, the current digital human live broadcast room mainly relies on pre-set scripts to introduce and explain the commodities in the live broadcast. However, for temporary situations that occur during the live broadcast, such as questions from users about unlisted commodities, the digital human anchor cannot respond accordingly, resulting in poor interactivity between the digital human anchor and the user, affecting the user's viewing experience. Therefore, how to update the live broadcast room script in real time according to the user's intention in the digital human live broadcast room is a problem that needs to be solved. Summary of the invention

[0003] The present disclosure provides a live broadcast control method, apparatus, device and storage medium.

[0004] According to one aspect of the present disclosure, a live broadcast control method is provided, comprising:

[0005] Obtain user interaction information in the live broadcast room;

[0006] Performing intent recognition on the user interaction information to obtain the user intent; the user intent includes the identifier of the first product that the user wants to explain, and the user's focus on the first product;

[0007] Obtain product details of the first product based on the user intention;

[0008] Generate a live broadcast script for the first product based on the product details and focus points;

[0009] Use the live broadcast script of the first product to modify the live broadcast room script and play the modified live broadcast room script.

[0010] According to another aspect of the present disclosure, a live broadcast control device is provided, comprising:

[0011] The first acquisition module is used to obtain user interaction information in the live broadcast room;

[0012] An intention recognition module is used to perform intention recognition on the user interaction information to obtain the user intention; the user intention includes the identifier of the first product that the user wants to explain, and the user's focus on the first product;

[0013] A second acquisition module, configured to acquire product details of the first product based on the user intention;

[0014] A generation module, used to generate a live broadcast script for the first product according to the product details and focus points;

[0015] The modification module is used to modify the live broadcast room script using the live broadcast script of the first product, and play the modified live broadcast room script.

[0016] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any method in the embodiments of the present disclosure.

[0020] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute any method according to the embodiments of the present disclosure.

[0021] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which implements any method according to the embodiments of the present disclosure when executed by a processor.

[0022] The present invention identifies user intentions based on user interaction information, generates a live broadcast script for the products that the user is interested in based on the user intentions, and then realizes the modification of the live broadcast script, which can enhance the user's sense of participation and experience.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0025] Figure 1 is a flow chart of an implementation of a live broadcast control method 100 according to an embodiment of the present disclosure;

[0026] Figure 2 is a flow chart of an implementation of a live broadcast script adjustment method 200 according to an embodiment of the present disclosure;

[0027] Figure 3 is a structural diagram of a live broadcast control device 300 according to an embodiment of the present disclosure;

[0028] Figure 4 is a structural schematic diagram of a live broadcast control device 400 according to an embodiment of the present disclosure;

[0029] Figure 5 A schematic block diagram of an example electronic device 500 that may be used to implement embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0030] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure 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 of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0031] The “and / or” in the embodiments of the present disclosure indicates that three relationships may exist. For example, A and / or B may indicate the following three situations: A exists alone, A and B exist at the same time, and B exists alone. The term “at least one” herein indicates any combination of at least two of any one or more of a plurality of types. For example, at least one of A, B, and C may indicate any one or more elements selected from the set consisting of A, B, and C. The terms “first” and “second” herein refer to and distinguish between multiple similar technical terms, and do not mean to limit the order or to limit the meaning to only two. For example, the first feature and the second feature refer to two types / two features. The first feature may be one or more, and the second feature may also be one or more.

[0032] In recent years, thanks to the rapid progress of artificial intelligence, the application of digital human technology in commodity live broadcasting scenarios has become increasingly mature. However, there are still certain limitations in the functional application of digital human live broadcast rooms, mainly manifested in that digital humans mainly focus on the introduction of a single product and lack the ability to flexibly respond to the diverse needs of users. Specifically, when users ask about unlisted products during the live broadcast, it is often difficult for digital humans to adjust the explanation content in real time, which limits their ability to provide users with a richer, more dynamic and personalized viewing and shopping experience. Therefore, how to optimize the live broadcast script in the digital human live broadcast room according to the user's intentions and needs is an urgent problem to be solved.

[0033] In order to solve the above problems, the embodiments of the present disclosure propose a live broadcast control method and device. Among them, the live broadcast control method and device proposed in the embodiments of the present disclosure can be applied to electronic devices such as mobile phones, computers, intelligent voice interaction devices, smart home appliances, car terminals, game consoles, e-book readers, multimedia playback devices, wearable devices, etc. The electronic device obtains user interaction information in the live broadcast room, and identifies the user interaction information to obtain the user intention. According to the user intention, the product details of the user's consultation product are obtained, and a live broadcast script for the consultation product is generated based on the product details. The live broadcast script of the product is used to modify the live broadcast script, and finally the modified live broadcast script is played. Alternatively, the live broadcast control method and device proposed in the embodiments of the present disclosure can also be applied to a cloud server. After obtaining the user interaction information, the user intention is identified, and the live broadcast script of the product is generated according to the user intention, and the live broadcast script is fed back to the terminal device. The terminal device modifies the live broadcast script according to the live script and plays the modified live broadcast script.

[0034] Figure 1 is a flow chart of an implementation of a live broadcast control method 100 according to an embodiment of the present disclosure, including:

[0035] S110, obtaining user interaction information in the live broadcast room;

[0036] S120, performing intent recognition on the user interaction information to obtain the user intent; the user intent includes the identifier of the first product that the user wants to explain, and the user's focus on the first product;

[0037] S130, obtaining product details of the first product based on the user's intention;

[0038] S140, generating a live broadcast script for the first product according to the product details and focus points;

[0039] S150. Use the live broadcast script of the first product to modify the live broadcast room script, and play the modified live broadcast room script.

[0040] In the disclosed embodiment, the user intent is identified to obtain the user intent consisting of the identifier of the first product and the user's focus on the first product, wherein the first product includes products that are not available in the live broadcast. The user intent is used to obtain detailed information of the first product, and based on the detailed information and focus of the first product, a live broadcast script explaining the first product is generated, and the live broadcast room is modified using the generated live broadcast script.

[0041] The live broadcast control method proposed in the embodiment of the present disclosure identifies user intentions based on user interaction information, generates a live broadcast script for the products that the user is interested in based on the user intentions, and then realizes the modification of the live broadcast room script. This method can enhance the user's sense of participation and experience.

[0042] In some implementations, performing intent recognition on user interaction information to obtain user intent includes:

[0043] Input each user interaction information into a pre-trained intention recognition model, and the intention recognition model outputs a first intention corresponding to each user interaction information; the first intention includes a product identifier and a focus of the product;

[0044] Based on the identifiers of the commodities included in the first intent, the multiple first intents are grouped to obtain multiple intent groups, each intent group includes multiple first intents, and each intent group corresponds to an identifier of a commodity;

[0045] For each intent group, when the number of first intents in the intent group is greater than or equal to a preset threshold, the user intent corresponding to the intent group is determined.

[0046] In the embodiment of the present disclosure, the intent recognition model can be obtained by pre-training, wherein the model training method is:

[0047] (1) Data collection: collect a large amount of user interaction information with user intent. This interaction information can come from actual live broadcast scenarios or be obtained through simulation methods. The collected user interaction information is labeled with intent, that is, the intent category expressed by each piece of interaction information is clearly defined. The intent category can be defined based on actual application scenarios, such as querying product information, making a request for product explanation, etc.

[0048] (2) Data preprocessing: remove irrelevant characters, emoticons, etc. in user interaction information to ensure the purity of text data. Perform word segmentation on the text data so that subsequent models can better understand and process the text data. Convert the segmented text into numerical form. Usually, word embedding technology can be used, such as the word text to vector conversion method (Word to Vector, Word2Vec), global vector word embedding (Global Vectors for Word Representation, GloVe) or bidirectional encoder representation method (Bidirectional Encoder Representations from Transformers, BERT), etc., to convert the text into a vector form that the model can process.

[0049] (3) Select a suitable intent recognition model based on task requirements and data characteristics. Common models include Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Transformer models, etc. At the same time, select a suitable model architecture, including input layer, hidden layer and output layer.

[0050] (4) Divide the labeled user interaction information data into training set, validation set and test set, and set the training parameters, including learning rate, batch size, training rounds, etc. Use the training set data to train the model, and optimize the model parameters through the back propagation algorithm so that the model can accurately identify the intent in the user interaction information. Use the validation set data to evaluate the model during the training process, and adjust the model parameters or structure according to the evaluation results to improve the performance of the model.

[0051] (5) Deploy the trained model to the actual application scenario, use the deployed model to identify the intent of user interaction information, and perform corresponding processing or response based on the recognition results.

[0052] In the embodiment of the present disclosure, the intent recognition model recognizes and outputs multiple corresponding first intents for multiple user interaction messages, and the first intent contains the identifiers of the goods, which are usually the product name, brand name, model, product alias, etc., used to refer to specific goods in the market. At the same time, the first intent incorporates the user's concerns about the goods, which may involve the performance characteristics of the goods (such as processor speed, screen resolution, etc.), functional characteristics (such as waterproof level, use effect, etc.), appearance design (such as color, material, etc.), price range, user evaluation and after-sales service, etc.

[0053] In the disclosed embodiment, first intents with the same product identifier will be grouped into the same group to form an intent group. Each intent group clearly points to a specific product and includes multiple focus points on the product.

[0054] For example, N user interaction information is input into the intent recognition model, and N first intentions are obtained after being processed by the intent recognition model. Among them, these N first intentions may cover different products and concerns. Among the N first intentions, they may include: product A+concern 1, product A+concern 2, product A+concern 3, product B+concern 1... In order to effectively utilize these first intentions, the N first intentions are grouped according to the product identifier mentioned in each first intention. Specifically, by traversing these N first intentions, the product identifier mentioned in each first intention is identified, and the first intentions with the same product identifier are grouped into the same group. According to this logic, all first intentions that mention product A (such as "product A+concern 1", "product A+concern 2", "product A+concern 3", etc.) will be grouped into one group, and the product identifier corresponding to this intention group is "product A".

[0055] Similarly, the remaining first intents are processed according to the same product identification grouping strategy, and finally M intent groups are obtained. Each intent group revolves around a specific product and contains multiple first intents. These intentions start from different angles or focus points and jointly describe the user's interest and demand for the product.

[0056] When further processing and analyzing the intent grouping driven by product identification, a preset threshold is introduced as a basis for judgment to ensure that the selected intent grouping can effectively reflect the user's actual intention. Specifically, for each intent grouping, the number of first intentions contained therein is counted and compared with the preset threshold. This comparison process determines which intent groups can be regarded as valid and important carriers of user intent. If the number of first intentions in an intent grouping is greater than or equal to the preset threshold, it means that the first intentions in the intent grouping are more concentrated and have a higher demand for a specific product. The user intent can be determined based on the corresponding first intention in the intent grouping.

[0057] In addition, the setting of the preset threshold needs to be flexibly adjusted according to the actual situation. On the one hand, a threshold that is too high may cause some valid intent groups that should have been identified to be missed; on the other hand, a threshold that is too low may introduce too much noise and irrelevant information, affecting the accuracy and effectiveness of determining user intent. Therefore, the present disclosure calculates the average value of the number of first intents in each group (that is, the ratio of the number of first intents to the number of intent groups) and uses this as a preliminary threshold. In order to more accurately reflect the concentration of user intentions, it is necessary to further analyze the distribution of the number of intents and adjust the preliminary threshold to set a preset threshold that can effectively filter noise data and accurately capture the real needs of users.

[0058] By adopting the above method, by inputting the user interaction information into the pre-trained intent recognition model, the model can parse the user's intent and group the intent according to the product's identification. Then, by setting a preset threshold, it is used to filter out the user intent groups that have enough (i.e., the number is greater than the preset threshold) first intents for a specific product, which is beneficial to accurately obtain the user's intention for the product, and then realize feedback to the user, thereby improving the user experience.

[0059] In some implementations, determining the user intent corresponding to the intent group includes:

[0060] Determine the identifier of the product corresponding to the intent group as the identifier of the first product that the user requires to be explained and that is included in the user intent;

[0061] The focus points of the products included in the multiple first intentions in the intention group are clustered to obtain multiple cluster center points, each cluster center point corresponds to a focus point, and the focus points corresponding to the multiple cluster center points are used as the user's focus points on the first product included in the user intent.

[0062] In the disclosed embodiment, the identifier of the commodity corresponding to the intent group is used as the identifier of the commodity that needs to be explained in the user's intent, that is, the identifier of the first commodity. When further exploring the details of the user's intent, it is necessary to conduct an in-depth analysis of the multiple first intentions in each intent group to obtain the categories of focus points in each intent group. These focus points may involve multiple aspects such as the characteristics, functions, and materials of the commodity. In order to systematically understand and summarize these focus points, the focus points are processed through a clustering algorithm.

[0063] During the clustering process, the algorithm will classify all the first intentions in the intent group according to the similarity between the focus points to form multiple clusters. The points within each cluster have a high degree of similarity, while the points between different clusters are relatively independent. After clustering is completed, each cluster will be abstracted into a cluster center point, which represents the core features of all the focus points in the cluster. For example, three users have the following focus points on the same pair of sneakers:

[0064] (1) User 1 asks how heavy this pair of sneakers is;

[0065] (2) User 2 asks how much the pair of sneakers weigh;

[0066] (3) User 3 asks whether this pair of sneakers is breathable.

[0067] For the above three focus points, the focus points of user 1 and user 2 can be clustered into one category, and the center point of the cluster is “the weight of sports shoes”. In addition, the focus points of user 3 can be clustered into one category, and the center point of the cluster is “the breathability of sports shoes”.

[0068] In the disclosed embodiment, the focus points corresponding to the multiple cluster centers can be identified as the focus points of the user on the first product in the user's intention. These focus points provide important basis for subsequent personalized recommendations, product introductions, question answers, etc.

[0069] By adopting the above method, by determining that the product identifier corresponding to the intent group is the identifier of the first product that the user requests to be explained, the live broadcast room can focus on the specific product that the user is interested in or needs to know about. Furthermore, cluster analysis is performed on the product focus points mentioned in multiple first intents in the intent group to form multiple clear focus point cluster centers, which can summarize the user's main interest points in the first product and improve the efficiency of understanding the user's intent.

[0070] In some implementations, obtaining product details of the first product based on the user intent includes:

[0071] Based on the identifier of the first product included in the user intention, product details of the first product are searched from a pre-stored product library.

[0072] In the disclosed embodiment, the product library is a large and orderly data set that stores detailed information of all products on sale on the platform, including product names, prices, specifications, etc. This information is organized and classified to ensure that it can be quickly retrieved and called when needed.

[0073] Based on the ID of the first product included in the user's intent, if the first product is not available in the current live broadcast, the system will execute a search algorithm in the product library, which can quickly traverse the entire database to find a product that matches the ID of the given first product. After finding a match, the system will further verify the validity and completeness of the product in order to extract accurate product details.

[0074] By adopting the above method, by quickly searching and obtaining detailed information of the product from a pre-saved product library based on the identifier of the first product included in the user's intention, the efficiency of searching for product details is improved, and the speed and convenience of obtaining information are improved.

[0075] In some implementations, generating a live broadcast script for the first product based on product details and focus points includes:

[0076] According to the focus point, extract the product introduction information corresponding to the focus point from the product details;

[0077] Determine the type of the first product according to the identifier of the first product; and search for a live broadcast script template corresponding to the type;

[0078] Generate a live broadcast script for the first product based on the product introduction information and the live broadcast script template corresponding to the type.

[0079] In the disclosed embodiment, according to the user's focus on the first product, the product introduction information matching the user's focus is extracted by analyzing various contents in the product details.

[0080] According to the identifier of the first commodity, the type of the first commodity is determined through a database or a commodity information management system. The type of commodity may cover multiple dimensions, such as clothing (such as tops, pants, cotton-padded clothes, etc.), electronics (such as mobile phones, computers, home appliances, etc.), and food (vegetables, fruits, fresh food, etc.). After determining the type of the first commodity, a live broadcast script template that matches the commodity type can be searched in a database for storing the correspondence between commodity types and live broadcast script templates.

[0081] Based on the introduction information of the first product and the live broadcast script template corresponding to the type of the first product, a live broadcast script for the first product is generated. The live broadcast script template provides a basic product explanation framework, and the live broadcast script template can be personalized according to the introduction information of the first product to make it closer to the actual live broadcast situation.

[0082] For example, in a live sales event, some users raised some concerns about a cotton and linen shirt, including "fabric material", "design features" and "color selection". Based on the above three concerns, the product introduction information corresponding to the concerns can be extracted from the product details:

[0083] (1) Fabric material: Made of high-quality cotton and linen blended fabric, light and breathable, comfortable to wear, suitable for spring and summer.

[0084] (2) Design features: Simple yet fashionable design, exquisitely tailored neckline and cuffs, smooth body, easy to match with various bottoms.

[0085] (3) Color selection: We offer a variety of fresh and elegant colors for you to choose from, including white, light blue, light pink, etc.

[0086] According to the label of "cotton and linen shirt", its type is determined to be "fashion clothing-shirt category". Then, the template corresponding to this type is searched in the live script template library. The template contains introductions to clothing fabrics, styles, matching suggestions, etc. Based on the extracted product introduction information and the found live script template, a live script about "cotton and linen shirt" can be generated.

[0087] By adopting the above method, the product introduction information corresponding to the focus point is extracted from the product details, and the live broadcast script template is searched according to the product type. Finally, the live broadcast script is generated according to the product introduction information, thereby improving the efficiency of live broadcast script generation.

[0088] In some implementations, the live broadcast room script includes live broadcast scripts for multiple commodities;

[0089] Use the live broadcast script of the first product to modify the live broadcast script, including:

[0090] Determine the attention level of the first product;

[0091] Determine the frequency of explaining the first product based on the attention paid to the first product;

[0092] According to the frequency of explanation of the first product, add the live broadcast script of the first product to the live broadcast room script.

[0093] In some implementations, determining the attention level of the first product includes:

[0094] Determine the user intent corresponding to the first product;

[0095] Determine an intent group corresponding to the user intent corresponding to the first product;

[0096] The number of first intentions included in the intention group is determined, and the attention level of the first product is determined according to the number of first intentions.

[0097] In the disclosed embodiment, the live broadcast script of the first product is used to modify the live broadcast room script. First, the attention of the first product must be determined. Regarding the attention of the first product, the user intent corresponding to the first product is first determined, and then the intent group corresponding to the user intent is determined, and the attention of the first product is determined according to the number of first intents in the intent group. If the number of first intents for the first product is larger, it means that the market has greater interest and demand for the product, and its attention is higher. Conversely, if the number of first intents is small, it may mean that the product has limited appeal in the market.

[0098] After determining the attention level of the first product, it is necessary to plan the frequency of its explanation in the live broadcast based on this attention level. If the attention level of the first product is high, then it should become the highlight of the live broadcast, and accordingly, the explanation frequency should be set relatively high; if the attention level is moderate, it can be explained interspersed with other products to ensure that the audience can maintain a certain attention to the first product while learning about other products; if the attention level is low, the explanation frequency can be appropriately set relatively low. After clarifying the explanation frequency of the first product, the display content and time arrangement of the live broadcast script of the first product will be planned in the live broadcast script based on this frequency, so as to modify the live broadcast script.

[0099] By adopting the above method, the frequency of explanation is determined according to the popularity of the first product, ensuring that the live broadcast content can be developed around the products that users are interested in, thereby increasing the attractiveness of the live broadcast. At the same time, adding live broadcast scripts for products that users are interested in, through detailed introduction and display, helps the audience to have a more comprehensive understanding of the product characteristics and improves the sales conversion rate.

[0100] In some embodiments, it further comprises:

[0101] Determine the commodity value of each commodity in the live broadcast room based on the live broadcast data;

[0102] Based on the commodity value of each commodity, determine the frequency of explaining the live broadcast script of each commodity;

[0103] Adjust the live broadcast script based on the frequency of explanations.

[0104] In the disclosed embodiment, in order to accurately grasp the value of each product in the live broadcast room, it is necessary to comprehensively collect and analyze various data during the live broadcast process. These data include the number of viewers, audience stay time, product click-through rate, conversion rate, sales, etc.

[0105] The number of viewers and the length of time viewers stay on the live broadcast can be used to preliminarily determine the popularity of the live broadcast room and the audience's interest in the live broadcast content, which indirectly reflects the attractiveness of the product to the audience. If a product can significantly attract viewers to stay and watch when it is displayed, then the potential value of the product in the live broadcast room is relatively high.

[0106] The click-through rate and conversion rate of products are direct indicators of the performance of products in the live broadcast room. The click-through rate reflects the audience's interest and curiosity in the product, while the conversion rate is the key to converting this interest into actual purchasing behavior. By comparing the click-through rate and conversion rate of different products, we can clearly see which products are more popular with the audience, thereby determining the value of the products.

[0107] Sales volume is also an important aspect that cannot be ignored in evaluating the value of goods. High sales volume not only represents the high sales volume of goods, but also reflects the popularity of goods in the market. Combining sales volume with cost data, we can calculate the profit margin of goods and further confirm their commercial value.

[0108] After evaluating the commodity value of each commodity in the live broadcast room, determine and adjust the explanation frequency of the live broadcast script of each commodity based on these values. For commodities with higher commodity value, the explanation frequency can be increased. High-value commodities often have unique selling points, wide market demand and high profit margins, so they should be given more exposure opportunities during the live broadcast. For commodities with moderate commodity value, flexibly adjust their explanation frequency according to their performance in the live broadcast room and market feedback, ensuring that it is neither too long nor too short, so as to fully demonstrate their characteristics and advantages while attracting the audience's attention. For commodities with lower commodity value or poor market performance, the explanation frequency will be appropriately reduced to avoid wasting live broadcast time and audience resources. Finally, adjust the live broadcast room script according to the explanation frequency of the live broadcast script.

[0109] Figure 2 It is a flow chart of the implementation of the live broadcast script adjustment method 200 according to an embodiment of the present disclosure.

[0110] like Figure 2 As shown, the live broadcast script adjustment method mainly includes four stages: live broadcast script and product information transmission, pre-broadcast strategy, on-broadcast strategy and on-broadcast architecture construction.

[0111] S210. Users and merchants transmit live broadcast scripts and information about products for sale.

[0112] S220, pre-broadcast strategy, used to determine the status of the live broadcast script, including:

[0113] S221, determining whether the live broadcast room script is added or modified;

[0114] S222. If the live broadcast room script has not been added or modified, the live broadcast room script is directly used; if the live broadcast room script has been added or modified, the on-air strategy is adopted to automatically adjust the script.

[0115] S230, a live broadcast strategy, for automatically adjusting the live broadcast script added or modified by the user, including adjusting the default explanation order and real-time adjustment, wherein adjusting the default explanation order includes the following steps:

[0116] S231, enter the cold start period of live broadcast, and continue to use the set live broadcast script. For example, the live broadcast script initially set by the user is: four products A, B, C, and D are explained once each. Among them, when the number of visits (Page View, PV) of the live broadcast room is less than the threshold (shown as x in the figure), it means that the live broadcast is still in the data accumulation stage, and the set live broadcast script is used in the live broadcast;

[0117] The PV threshold x can be set based on historical data. Knowing the average visitor level of the live broadcast of the product being sold can help set a threshold that is neither conservative nor too aggressive.

[0118] S232. When PV is greater than x, the live broadcast room script adjustment period begins, that is, the new product ranking and frequency are determined based on the value of the products. This process is essentially to adjust the frequency of explanation of the live broadcast scripts of each product. Among them, high-value products may give their live broadcast scripts a higher frequency of explanation, that is, more time is arranged to explain the live broadcast script of the product during the entire live broadcast. Continuing with the previous example, through value evaluation, the ratio of the number of explanations of the four products A, B, C, and D is 3:2:1:0, that is, A has the highest value, and the frequency of explanation of A's live broadcast script is the highest during the live broadcast; B and C have medium values, and the frequency of explanation of B and C's live broadcast scripts is medium during the live broadcast; D has the lowest value, and D's live broadcast script is not explained during the live broadcast.

[0119] Real-time product adjustment includes the following steps:

[0120] S233, the user requests an explanation of a certain product, for example, the user comments: Let me talk about product D;

[0121] S234, judging whether the number of people in the current live broadcast room is less than the number threshold (shown as n in the figure);

[0122] S235. If the number of people in the current live broadcast room is less than n, insert the live broadcast script for the product requested by the user in real time. After explaining the requested product, continue to broadcast according to the adjusted default live broadcast room script. Continuing with the previous example, after explaining A, insert the live broadcast script for D in real time. After explaining D, continue to broadcast according to the adjusted live broadcast room script;

[0123] S236. If the number of people in the live broadcast room is greater than n, the default explanation order is used.

[0124] When there are fewer people in the live broadcast room, each user's request may have a greater impact on the live broadcast content. Therefore, if the number of people in the live broadcast room is less than the threshold n, the digital human anchor can flexibly respond to the user's request and insert the explanation of the user's requested product in real time. When there are more people in the live broadcast room, if every user's request is responded to in real time, it may cause the live broadcast content to become disorganized, affecting the overall rhythm and efficiency of the live broadcast. Therefore, setting the threshold n can be used as a boundary. When the number of people in the live broadcast room exceeds this limit, the digital human anchor can continue to use the default explanation order to ensure the consistency and stability of the live broadcast content.

[0125] The number threshold n can be adjusted accordingly based on the size of the live broadcast room. For example, for a small live broadcast room, since the audience base is relatively small, each user request may have a greater impact on the live broadcast content. Therefore, the threshold n can be appropriately increased to better meet user needs. For a large live broadcast room, since the number of viewers is large, frequent responses to user requests may disrupt the live broadcast rhythm, so the threshold n can be appropriately lowered.

[0126] S240. According to the adjustment strategy of the live broadcast room script, adjust the structure of the live broadcast room script during the broadcast:

[0127] S241. According to the example of S232, after adjusting the explanation frequency of the live broadcast script of the product according to the value of the product, the live broadcast script can be adjusted to: ABABAC;

[0128] S242, according to the example of S235, combined with the default live broadcast room script adjusted in step S241, after inserting the live broadcast script for D in real time, the live broadcast room script can be updated in real time to: ADBABAC;

[0129] S243. After explaining D's live broadcast script, restore the adjusted default live broadcast script: ABABAC.

[0130] By using the above method, the value of each commodity in the live broadcast room is determined by combining the live broadcast data, and the explanation frequency of each commodity in the live broadcast script is adjusted accordingly, which can ensure that high-value commodities receive more exposure and explanation time, thereby improving the pertinence and effectiveness of the explanation. At the same time, it can also provide feedback on users' product explanation requests, thereby improving users' experience of watching live broadcasts.

[0131] The disclosed embodiment also provides a live broadcast control device. Figure 3 3 is a structural diagram of a live broadcast control device 300 according to an embodiment of the present disclosure, including:

[0132] The first acquisition module 310 is used to acquire user interaction information in the live broadcast room;

[0133] The intention recognition module 320 is used to perform intention recognition on the user interaction information to obtain the user intention; the user intention includes the identifier of the first product that the user wants to explain, and the user's focus on the first product;

[0134] A second acquisition module 330, configured to acquire product details of the first product based on the user intention;

[0135] A generating module 340, configured to generate a live broadcast script of the first product according to the product details and focus points;

[0136] The modification module 350 is used to modify the live broadcast room script using the live broadcast script of the first product, and play the modified live broadcast room script.

[0137] In some implementations, the intent recognition module 320 is used to:

[0138] Input each user interaction information into a pre-trained intent recognition model, and the intent recognition model outputs a first intent corresponding to each user interaction information; the first intent includes a product identifier and a focus of the product;

[0139] Based on the identifiers of the commodities included in the first intent, the multiple first intents are grouped to obtain multiple intent groups, each intent group includes multiple first intents, and each intent group corresponds to an identifier of a commodity;

[0140] For each intent group, when the number of first intents in the intent group is greater than or equal to a preset threshold, the user intent corresponding to the intent group is determined.

[0141] In some implementations, the intent recognition module 320 is used to:

[0142] Determine the identifier of the product corresponding to the intent group as the identifier of the first product that the user requires to be explained and that is included in the user intent;

[0143] The focus points of the products included in the multiple first intentions in the intention group are clustered to obtain multiple cluster center points, each cluster center point corresponds to a focus point, and the focus points corresponding to the multiple cluster center points are used as the user's focus points on the first product included in the user intent.

[0144] In some implementations, the second acquisition module 330 is configured to search for product details of the first product from a pre-stored product library based on the identifier of the first product included in the user intention.

[0145] In some implementations, the generation module 340 is used to:

[0146] According to the focus point, extract the product introduction information corresponding to the focus point from the product details;

[0147] Determine the type of the first product according to the identifier of the first product; and search for a live broadcast script template corresponding to the type;

[0148] Generate a live broadcast script for the first product based on the product introduction information and the live broadcast script template corresponding to the type.

[0149] In some implementations, the live broadcast room script includes live broadcast scripts for multiple commodities;

[0150] The modification module 350 is used to:

[0151] Determine the attention level of the first product;

[0152] Determine the frequency of explaining the first product based on the attention paid to the first product;

[0153] According to the frequency of explanation of the first product, add the live broadcast script of the first product to the live broadcast room script.

[0154] In some embodiments, the modification module 350 is used to:

[0155] Determine the user intent corresponding to the first product;

[0156] Determine an intent group corresponding to the user intent corresponding to the first product;

[0157] The number of first intentions included in the intention group is determined, and the attention level of the first product is determined according to the number of first intentions.

[0158] Figure 4 is a structural diagram of a live broadcast control device 400 according to an embodiment of the present disclosure, which, in some implementations, further includes:

[0159] A first determination module 460 is used to determine the commodity value of each commodity in the live broadcast room according to the live broadcast data;

[0160] The second determination module 470 is used to determine the explanation frequency of the live broadcast script of each commodity based on the commodity value of each commodity;

[0161] The adjustment module 480 is used to adjust the live broadcast script based on the explanation frequency. The specific functions and examples of each module and submodule of the device of the embodiment of the present disclosure can be found in the relevant description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0162] In the technical solution disclosed in the present invention, the acquisition, storage and application of personal information of users involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0163] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0164] Figure 5A schematic block diagram of an example electronic device 500 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0165] like Figure 5 As shown, the device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0166] A number of components in the device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0167] The computing unit 501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 501 performs the various methods and processes described above, such as detection methods. For example, in some embodiments, the detection method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of the detection method described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the detection method in any other appropriate manner (e.g., by means of firmware).

[0168] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0169] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0170] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 foregoing.

[0171] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0172] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0173] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0174] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0175] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A live broadcast control method, comprising: Obtain user interaction information in the live broadcast room; Performing intent recognition on the user interaction information to obtain the user intent; The user intention includes the identifier of the first product that the user wants to explain, and the user's focus on the first product; Acquire product details of the first product based on the user intention; Generate a live broadcast script for the first product according to the product details and the focus points; Use the live broadcast script of the first product to modify the live broadcast room script and play the modified live broadcast room script.

2. The method according to claim 1, wherein: The performing intention recognition on the user interaction information to obtain the user intention includes: Input each of the user interaction information into a pre-trained intention recognition model, and the intention recognition model outputs a first intention corresponding to each of the user interaction information; the first intention includes a product identifier and a focus of the product; Based on the identifiers of the commodities included in the first intentions, grouping the plurality of the first intentions to obtain a plurality of intention groups, each of the intention groups including a plurality of first intentions, and each of the intention groups corresponding to an identifier of a commodity; For each of the intent groups, when the number of first intents in the intent group is greater than or equal to a preset threshold, the user intent corresponding to the intent group is determined.

3. The method according to claim 2, wherein: The determining the user intent corresponding to the intent group includes: Determine the identifier of the commodity corresponding to the intention group as the identifier of the first commodity that the user requires to be explained and is included in the user intention; The focus points of the commodities contained in the multiple first intentions in the intention group are clustered to obtain multiple cluster center points, each of the cluster center points corresponds to a focus point, and the focus points corresponding to the multiple cluster center points are used as the user's focus points on the first commodity contained in the user intent.

4. The method according to any one of claims 1 to 3, wherein: The acquiring the product details of the first product based on the user intention includes: Based on the identifier of the first product included in the user intention, product details of the first product are searched from a pre-stored product library.

5. The method according to any one of claims 1 to 3, wherein: The generating a live broadcast script of the first product according to the product details and the focus point includes: According to the focus point, extracting product introduction information corresponding to the focus point from the product details; Determine the type of the first commodity according to the identifier of the first commodity; and search for a live broadcast script template corresponding to the type; Based on the product introduction information and the live broadcast script template corresponding to the type, generate a live broadcast script for the first product.

6. The method according to claim 2 or 3, wherein: The live broadcast room script includes live broadcast scripts for multiple commodities; The method of using the live broadcast script of the first product to modify the live broadcast room script includes: Determining the attention level of the first product; Determining a frequency of explaining the first product according to the attention level of the first product; According to the frequency of explaining the first product, a live broadcast script for the first product is added to the live broadcast room script.

7. The method according to claim 6, wherein: The determining the attention level of the first commodity includes: Determining a user intention corresponding to the first product; Determine an intent group corresponding to the user intent corresponding to the first product; The number of first intentions included in the intention group is determined, and the attention level of the first product is determined according to the number of the first intentions.

8. The method according to claim 6, further comprising: Determine the commodity value of each commodity in the live broadcast room according to the live broadcast data; Based on the commodity value of each commodity, determine the explanation frequency of the live broadcast script of each commodity; The live broadcast script is adjusted based on the explanation frequency.

9. A live broadcast control device, comprising: The first acquisition module is used to obtain user interaction information in the live broadcast room; An intention recognition module, used to perform intention recognition on the user interaction information to obtain the user intention; The user intention includes the identifier of the first product that the user wants to explain, and the user's focus on the first product; A second acquisition module, configured to acquire product details of the first product based on the user intention; A generating module, configured to generate a live broadcast script for the first product according to the product details and the focus point; The modification module is used to modify the live broadcast room script using the live broadcast script of the first product, and play the modified live broadcast room script.

10. The device according to claim 9, wherein: The intention recognition module is used to: Input each of the user interaction information into a pre-trained intention recognition model, and the intention recognition model outputs a first intention corresponding to each of the user interaction information; the first intention includes a product identifier and a focus of the product; Based on the identifiers of the commodities included in the first intentions, grouping the plurality of the first intentions to obtain a plurality of intention groups, each of the intention groups including a plurality of first intentions, and each of the intention groups corresponding to an identifier of a commodity; For each of the intent groups, when the number of first intents in the intent group is greater than or equal to a preset threshold, the user intent corresponding to the intent group is determined.

11. The device according to claim 10, wherein: The intention recognition module is used to: Determine the identifier of the commodity corresponding to the intention group as the identifier of the first commodity that the user requires to be explained and is included in the user intention; The focus points of the commodities contained in the multiple first intentions in the intention group are clustered to obtain multiple cluster center points, each of the cluster center points corresponds to a focus point, and the focus points corresponding to the multiple cluster center points are used as the user's focus points on the first commodity contained in the user intent.

12. The device according to any one of claims 9 to 11, wherein: The second acquisition module is used to search for product details of the first product from a pre-stored product library based on the identifier of the first product included in the user intention.

13. The device according to any one of claims 9 to 11, wherein: The generation module is used for: According to the focus point, extracting product introduction information corresponding to the focus point from the product details; Determine the type of the first commodity according to the identifier of the first commodity; and search for a live broadcast script template corresponding to the type; Based on the product introduction information and the live broadcast script template corresponding to the type, generate a live broadcast script for the first product.

14. The device according to claim 10 or 11, wherein: The live broadcast room script includes live broadcast scripts for multiple commodities; The modification module is used to: Determining the attention level of the first product; Determining a frequency of explaining the first product according to the attention level of the first product; According to the frequency of explaining the first product, a live broadcast script for the first product is added to the live broadcast room script.

15. The device according to claim 14, wherein: The modification module is used to: Determining a user intention corresponding to the first product; Determine an intent group corresponding to the user intent corresponding to the first product; The number of first intentions included in the intention group is determined, and the attention level of the first product is determined according to the number of the first intentions.

16. The apparatus according to claim 14, further comprising: A first determination module, used to determine the commodity value of each commodity in the live broadcast room according to the live broadcast data; A second determination module is used to determine the explanation frequency of the live broadcast script of each of the commodities based on the commodity value of each of the commodities; An adjustment module is used to adjust the live broadcast script based on the explanation frequency.

17. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

18. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-8.

19. A computer program product, comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 8.

Images