Item recommendation method, device, equipment and readable storage medium

Abstract

The present invention provides an item recommendation method, device, equipment and readable storage medium, and relates to the field of information technology of the Internet of Things. The item recommendation method includes determining a multidimensional interest matrix of a family user based on the historical viewing records of the family user, and determining the representation vector of each item; establishing a time period matrix corresponding to the multidimensional interest matrix; calculating the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multidimensional interest matrix; calculating the fusion weight based on the time period interest weight corresponding to the time period matrix and the real-time interest weight; calculating the real-time interest vector of the family user based on the fusion weight and the multidimensional interest matrix; and determining the target recommended item based on the real-time interest vector and the representation vector of the candidate item.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things information technology, and in particular to an item recommendation method, device, equipment and readable storage medium. Background Art

[0002] As people's living standards improve, various film and television resources continue to increase. Users need to spend a lot of time to find the videos they want to watch. Therefore, it is necessary to use personalized video recommendation methods to recommend videos that users are interested in. For family users, a family includes multiple family members, and each member generally has different interests and preferences. However, the current video recommendation cannot recommend suitable videos for family users. Therefore, the flexibility of existing item recommendations is poor. Summary of the invention

[0003] Embodiments of the present invention provide an item recommendation method, device, equipment and readable storage medium to solve the problem of poor flexibility in existing item recommendations.

[0004] To solve the above technical problems, the present invention is achieved as follows:

[0005] In a first aspect, an embodiment of the present invention provides an item recommendation method, comprising:

[0006] Determine a multi-dimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item;

[0007] Establishing a time period matrix corresponding to the multidimensional interest matrix;

[0008] Calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix;

[0009] A fusion weight is calculated based on the time period interest weight corresponding to the time period matrix and the real-time interest weight;

[0010] Calculate the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix;

[0011] A target recommended item is determined based on the real-time interest vector and the representation vector of the candidate item.

[0012] In a second aspect, an embodiment of the present invention provides an item recommendation device, comprising:

[0013] A first unit is used to determine a multi-dimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item;

[0014] The second unit is used to establish a time period matrix corresponding to the multi-dimensional interest matrix;

[0015] A third unit is used to calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix;

[0016] A fourth unit is used to calculate a fusion weight based on the time period interest weight corresponding to the time period matrix and the real-time interest weight;

[0017] A fifth unit, configured to calculate a real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix;

[0018] The sixth unit is used to determine a target recommended item based on the real-time interest vector and the representation vector of the candidate item.

[0019] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the item recommendation method described in the first aspect.

[0020] In a fourth aspect, an embodiment of the present invention provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the item recommendation method described in the first aspect are implemented.

[0021] In the embodiment of the present invention, a multidimensional interest matrix of family users is established, and the real-time interest weight of the family users is calculated based on the item representation vector corresponding to the real-time playback information of the family users and the multidimensional interest matrix, and the time period interest weight and the real-time interest weight corresponding to the multidimensional interest matrix are fused to obtain the fusion weight, and the real-time interest vector of the family users is determined based on the fusion weight and the multidimensional interest matrix, and the target recommended item is determined based on the real-time interest vector and the representation vector of the candidate item. In this way, when recommending items, the real-time interest of the family users is taken into consideration, which can solve the problem of poor flexibility of existing item recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0023] Figure 1 is a flow chart of an item recommendation method provided by an embodiment of the present invention;

[0024] Figure 2 is a module diagram of an item recommendation device provided by an embodiment of the present invention;

[0025] Figure 3 is a schematic diagram of a module structure of an electronic device provided by an embodiment of the present invention;

[0026] Figure 4 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] The item recommendation method provided in the embodiment of the present application is applicable to Over The Top (OTT) television and can be variably applied to adaptive recommendations in other application scenarios.

[0029] In this implementation, the item may refer to a video or a commodity. Taking video as an example, the recommended result may be different types of cartoon videos or TV series. Since the preferences of family members are different, teenagers prefer cartoon videos, and parents prefer TV series, it is necessary to recommend suitable items for different family members.

[0030] See also Figure 1 , Figure 1 is a flow chart of an item recommendation method provided by an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0031] Step 101: Determine a multi-dimensional interest matrix of a family user based on the family user's historical viewing records, and determine a representation vector for each item.

[0032] In this embodiment, the historical viewing record can be a record of the behavior of the family user in a certain period of time in the past, which can be a period of time one month ago, and the duration can be 10 days. This is only an example and not limited here. In other feasible embodiments, the selected time period can be adjusted within a certain range, but no matter how it is changed, it is within the scope of protection of the embodiment of the present application. In addition, the items in the historical viewing record are represented by vectors, which can facilitate subsequent calculations of each item.

[0033] Step 102: Establish a time period matrix corresponding to the multi-dimensional interest matrix.

[0034] In this implementation, the time period matrix corresponding to the multi-dimensional interest matrix can reflect the pattern of changes in the interests of home users over time.

[0035] Step 103: Calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix.

[0036] The real-time playback information of home users refers to the playback information close to the current playback time. For example, it can be the playback information of the day before the current playback time. This is only an example and is not limited to this. In other feasible implementations, the selected real-time time period can also be adjusted within a certain range.

[0037] Step 104, calculating a fusion weight based on the time period interest weight and the real-time interest weight corresponding to the time period matrix;

[0038] Step 105: Calculate the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix;

[0039] Step 106: Determine the target recommended item based on the real-time interest vector and the representation vector of the candidate item.

[0040] The candidate items may be videos that can be played or commodities that can be pushed in the entire network. In this implementation, a video that can be played is taken as an example for illustration.

[0041] The above-mentioned item recommendation method establishes a multi-dimensional interest matrix of family users, and calculates the real-time interest weight of family users based on the item representation vector corresponding to the real-time playback information of family users and the multi-dimensional interest matrix, and fuses the time period interest weight and the real-time interest weight corresponding to the multi-dimensional interest matrix to obtain the fusion weight, determines the real-time interest vector of family users based on the fusion weight and the multi-dimensional interest matrix, and determines the target recommended item based on the real-time interest vector and the representation vector of the candidate item. In this way, when recommending items, the real-time interest of family users is taken into consideration, which can solve the problem of poor flexibility of existing item recommendations.

[0042] Optionally, a multi-dimensional interest matrix of the family user is determined based on the family user's historical viewing records, and a representation vector of each item is determined, including:

[0043] Obtain the historical viewing records of the family user and the feature data corresponding to the historical viewing records.

[0044] Specifically, the feature data may be information associated with the object, and the information may be text used to describe the subject type of the object, or text representing the plot of the object, which is only used as an example and is not limited thereto.

[0045] Generate a multi-dimensional interest matrix of family users based on historical viewing records.

[0046] In this embodiment, the vector of items can be represented by I, and each item has a corresponding vector that satisfies the following formula.

[0047] I j ∈R m ;

[0048] Where m is the dimension of the vector, R is a real number, and j is the jth item. The representation vector corresponding to the item is generated based on the feature data corresponding to the historical viewing records.

[0049] It should be noted that when generating the vector of an item, the vector of the item can be determined based on the co-occurrence relationship of the item. Specifically, the item2vec algorithm can be used for generation, which will not be described in detail here.

[0050] In this embodiment, based on the feature data corresponding to the historical viewing records, the user's viewing behavior is represented as a multidimensional interest matrix K. The multidimensional interest matrix K is an n*m matrix, that is, it is composed of n m-dimensional vectors, n is the number of interests of the household user, and m is the same as the vector dimension of the object.

[0051] Specifically, when generating the multidimensional interest matrix K, the following method can be used, for example, by using the historical sequence clustering method, that is, first determining the number of n, and then clustering the items into n categories based on the generated item vectors. Then, for the home user h, the calculation method of the n-category interest vector can satisfy the following formula:

[0052]

[0053] Among them I j is the vector representation of the jth item, where j is an item in the user's history. n After the calculation is completed, z is the total number of items, and K is:

[0054]

[0055] Among them, each row of the multidimensional interest matrix K represents an item.

[0056] Optionally, a time period matrix corresponding to the multi-dimensional interest matrix is ​​established, including:

[0057] Obtain the time corresponding to the historical viewing record and determine the time period;

[0058] In this implementation, the time corresponding to the historical viewing record is transformed according to the period T0. It should be noted that the specific value of the period T0 can be one week or two weeks, and can be adjusted according to different application scenarios. This is only an example and not limited.

[0059] The cycle is divided into several time periods, and the items corresponding to the historical viewing records in the same time period are aggregated to obtain the historical viewing records corresponding to different time periods;

[0060] Furthermore, in this embodiment, the cycle T0 is divided into P time periods, and the items corresponding to the historical viewing records in the same time period are aggregated. For example, a cycle T0 is a week, which is further divided into 7 time periods, each time period is a day, and the items in the same day are aggregated, so that the historical viewing records of the family user in each of the seven days of the week can be obtained. Based on the historical viewing records of different time periods and the multidimensional interest matrix, a time period matrix corresponding to the multidimensional interest matrix is ​​established.

[0061] When the historical viewing records are transformed according to the period T0, the transformation method can be implemented by the following formula, for example.

[0062] t = t0 - n × T0;

[0063] Among them, t0 is the historical record time point before the transformation, and t is the historical record time point after the transformation. n∈N, after the transformation, t∈[0,T0).

[0064] Furthermore, based on the transformed historical record time points of the user and the multi-dimensional interest matrix generated above, the historical time period interest matrix T of the family user is generated, where T is a p*n dimensional matrix.

[0065] In a feasible implementation, a typical method of generating a time period interest matrix T is to calculate similarity between the aggregation results of items in different time periods p and the multidimensional interest matrix, and normalize the similarity as the historical interest vector weight of time period p.

[0066] Optionally, calculating the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix includes:

[0067] Acquire the real-time playback information of the home user, and calculate the representation vector of the item in the real-time playback information;

[0068] Calculating a real-time interest representation vector of a family user based on the representation vectors of the items in the real-time playback information;

[0069] The real-time interest weight of the family user is calculated based on the real-time interest representation vector of the family user and the multi-dimensional interest matrix.

[0070] In this optional implementation, the real-time playback information of the home user is first obtained, wherein the real-time playback information of the home user may refer to the playback information of a time period that is relatively close to the current playback time. For example, if the current playback time is 7 o'clock in the evening of the same day, then the playback information of the morning of the same day or the playback information of around 7 o'clock in the evening of the previous day may be obtained as the above-mentioned real-time playback information.

[0071] Furthermore, the item vector in the user's real-time playback information is determined, and the user's real-time interest representation vector is calculated based on the item vector, and the calculation method thereof may satisfy the following formula.

[0072]

[0073] In the formula, I R is the user's real-time interest representation vector, f(Δt) is the time decay function of real-time behavior, and a typical time decay function is e -Δt , I l is the representation vector of the item in the real-time playback information, r is the total number of item vectors, and l is the lth item vector.

[0074] Furthermore, the weight w of the real-time interest of the family user is calculated based on the user's real-time interest representation vector. R , where w R is an n-dimensional vector. Its calculation formula can satisfy the following relationship.

[0075]

[0076] In the formula, m k is the m-th dimension multidimensional interest matrix.

[0077] Optionally, the fusion weight is calculated based on the time period interest weight and the real-time interest weight corresponding to the time period matrix, including:

[0078] Set adjustment parameters;

[0079] The fusion weight is calculated based on the adjustment parameter, the time period interest weight corresponding to the time period matrix, and the real-time interest weight;

[0080] The adjustment parameter is used to indicate the ratio of the real-time interest to the historical interest of the family user.

[0081] In this embodiment, the adjustment parameter q can be set to adjust the proportion of real-time interest and historical interest. The larger the q, the more inclined to real-time interest. By setting the adjustment parameter q of weight fusion, the fusion ratio of real-time interest and historical period interest can be flexibly controlled.

[0082] The calculation formula for calculating the fusion weight can satisfy the following relationship.

[0083]

[0084] In the formula, w f is the fusion weight, T p It is the real-time interest weight at the current time.

[0085] Furthermore, the user's real-time interest vector is calculated based on the fusion weight and the multi-dimensional interest matrix of the family user. In this embodiment, its calculation formula can satisfy the following relationship.

[0086] u=w f ×K;

[0087] Where u is the user's real-time interest vector, where u is an m-dimensional vector.

[0088] Optionally, determining a target recommended item based on the real-time interest vector and the representation vector of the candidate item includes:

[0089] The candidate items corresponding to the similarity exceeding the similarity threshold are selected as the target recommended items.

[0090] In this implementation, the similarity between the real-time interest vector and the representation vector of the candidate item is first calculated, and the similarity is calculated in a manner that satisfies the following relationship.

[0091]

[0092] In the formula, m is the dimension, i is the i-th vector, and x i is the i-th real-time interest vector, y i is the representation vector of the i-th candidate item. In the above formula, the lower the value of sim, the higher the similarity.

[0093] Furthermore, candidate items corresponding to similarities exceeding a similarity threshold are selected as target recommended items.

[0094] The above-mentioned item recommendation method regards the user's real-time feedback information as the real-time interest weight w R , and represents the historical period interest as the historical period interest matrix T. A method is designed to combine the historical period interest with the real-time interest weight to calculate the user's real-time interest weight w f , can adaptively identify current family user members based on the user's real-time behavior, improving the flexibility of item recommendations.

[0095] See also Figure 2 . Figure 2 The present embodiment provides an item recommendation device 200, comprising:

[0096] The first unit 201 is used to determine a multi-dimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item;

[0097] The second unit 202 is used to establish a time period matrix corresponding to the multi-dimensional interest matrix;

[0098] The third unit 203 is used to calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix;

[0099] The fourth unit 204 is configured to calculate a fusion weight based on the time period interest weight corresponding to the time period matrix and the real-time interest weight;

[0100] The fifth unit 205 is used to calculate the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix;

[0101] The sixth unit 206 is configured to determine a target recommended item based on the real-time interest vector and the representation vector of the candidate item.

[0102] Optionally, the first unit 201 is specifically configured to:

[0103] Acquire historical viewing records of the family user and feature data corresponding to the historical viewing records;

[0104] generating a multi-dimensional interest matrix of the family user based on the historical viewing records;

[0105] A representation vector corresponding to the item is generated based on the feature data corresponding to the historical viewing record.

[0106] Optionally, the second unit 202 is specifically configured to:

[0107] Obtaining the time corresponding to the historical viewing record, and determining the period of the time;

[0108] Divide the cycle into a number of time periods, and aggregate the items corresponding to the historical viewing records in the same time period to obtain the historical viewing records corresponding to different time periods;

[0109] A time period matrix corresponding to the multidimensional interest matrix is ​​established based on the historical viewing records of the different time periods and the multidimensional interest matrix.

[0110] Optionally, the third unit 203 is specifically configured to:

[0111] Acquire the real-time playback information of the family user, and calculate the representation vector of the object in the real-time playback information;

[0112] Calculating the real-time interest representation vector of the family user based on the representation vector of the item in the real-time broadcast information;

[0113] The real-time interest weight of the family user is calculated according to the real-time interest representation vector of the family user and the multi-dimensional interest matrix.

[0114] Optionally, the fourth unit 204 is specifically configured to:

[0115] Set adjustment parameters;

[0116] The fusion weight is calculated based on the adjustment parameter, the time period interest weight corresponding to the time period matrix, and the real-time interest weight;

[0117] The adjustment parameter is used to indicate the ratio of the real-time interest to the historical interest of the family user.

[0118] Optionally, the sixth unit 206 is specifically configured to:

[0119] The candidate items corresponding to the similarity exceeding the similarity threshold are selected as the target recommended items.

[0120] The item recommendation device in the embodiment of the present application can be a device, or a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. Exemplarily, the mobile electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. The non-mobile electronic device can be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine or a self-service machine, etc., which is not specifically limited in the embodiment of the present application.

[0121] The item recommendation device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0122] The item recommendation device provided in the embodiment of the present application can achieve Figure 1 To avoid repetition, the various processes implemented in the method embodiment will not be described again here.

[0123] Optional, such as Figure 3 As shown, an embodiment of the present application also provides an electronic device 300, including a processor 301, a memory 302, and a program or instruction stored in the memory 302 and executable on the processor 301. When the program or instruction is executed by the processor 301, each process of the above-mentioned method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0124] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0125] Figure 4 A schematic diagram of the hardware structure of an electronic device to implement an embodiment of the present application.

[0126] The electronic device 400 includes but is not limited to components such as a radio frequency unit 401 , a network module 402 , an audio output unit 403 , an input unit 404 , a sensor 405 , a display unit 406 , a user input unit 407 , an interface unit 408 , a memory 409 , and a processor 410 .

[0127] Those skilled in the art will appreciate that the electronic device 400 may also include a power source (such as a battery) for supplying power to each component, and the power source may be logically connected to the processor 410 through a power management system, thereby implementing functions such as managing charging, discharging, and power consumption management through the power management system. Figure 4 The electronic device structure shown in the figure does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently, which will not be described in detail here.

[0128] The processor 410 is used for:

[0129] Determine a multi-dimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item;

[0130] Establishing a time period matrix corresponding to the multidimensional interest matrix;

[0131] Calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix;

[0132] A fusion weight is calculated based on the time period interest weight corresponding to the time period matrix and the real-time interest weight;

[0133] Calculate the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix;

[0134] A target recommended item is determined based on the real-time interest vector and the representation vector of the candidate item.

[0135] The processor 410 is further configured to:

[0136] Acquire historical viewing records of the family user and feature data corresponding to the historical viewing records;

[0137] generating a multi-dimensional interest matrix of the family user based on the historical viewing records;

[0138] A representation vector corresponding to the item is generated based on the feature data corresponding to the historical viewing record.

[0139] The processor 410 is further configured to:

[0140] Obtaining the time corresponding to the historical viewing record, and determining the period of the time;

[0141] Divide the cycle into a number of time periods, and aggregate the items corresponding to the historical viewing records in the same time period to obtain the historical viewing records corresponding to different time periods;

[0142] A time period matrix corresponding to the multidimensional interest matrix is ​​established based on the historical viewing records of the different time periods and the multidimensional interest matrix.

[0143] The processor 410 is further configured to:

[0144] Acquire the real-time playback information of the family user, and calculate the representation vector of the object in the real-time playback information;

[0145] Calculating the real-time interest representation vector of the family user based on the representation vector of the item in the real-time broadcast information;

[0146] The real-time interest weight of the family user is calculated according to the real-time interest representation vector of the family user and the multi-dimensional interest matrix.

[0147] The processor 410 is further configured to:

[0148] Set adjustment parameters;

[0149] The fusion weight is calculated based on the adjustment parameter, the time period interest weight corresponding to the time period matrix, and the real-time interest weight;

[0150] The adjustment parameter is used to indicate the ratio of the real-time interest to the historical interest of the family user.

[0151] The processor 410 is further configured to:

[0152] The candidate items corresponding to the similarity exceeding the similarity threshold are selected as the target recommended items.

[0153] In the embodiment of the present application, the electronic device 400 establishes a multi-dimensional interest matrix of the family user, and calculates the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix, fuses the time period interest weight and the real-time interest weight corresponding to the multi-dimensional interest matrix to obtain the fusion weight, determines the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix, and determines the target recommended item based on the real-time interest vector and the representation vector of the candidate item. In this way, when recommending items, the real-time interest of the family user is taken into consideration, which can solve the problem of poor flexibility of existing item recommendations.

[0154] It should be understood that in the embodiment of the present application, the input unit 404 may include a graphics processor (GPU) 4041 and a microphone 4042, and the graphics processor 4041 processes the image data of a static picture or video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 406 may include a display panel 4061, and the display panel 4061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 407 includes a touch panel 4071 and other input devices 4072. The touch panel 4071 is also called a touch screen. The touch panel 4071 may include two parts: a touch detection device and a touch controller. Other input devices 4072 may include, but are not limited to, a physical keyboard, a function key (such as a volume control button, a switch button, etc.), a trackball, a mouse, and a joystick, which will not be repeated here. The memory 409 may be used to store software programs and various data, including but not limited to applications and operating systems. The processor 410 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, and applications, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 410.

[0155] The embodiment of the present application also provides a readable storage medium on which a program or instruction is stored. When the program or instruction is executed by a processor, the above Figure 1 The various processes of the method embodiment can achieve the same technical effect, and to avoid repetition, they will not be described here.

[0156] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0157] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. An item recommendation method, characterized in that: include: Determine a multidimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item, wherein the multidimensional interest matrix is ​​used to represent the viewing behavior of the user, and the multidimensional interest matrix is ​​a matrix composed of n m-dimensional vectors, wherein n is the number of interests of the family user, m is the vector dimension of the item, and each row of the multidimensional interest matrix represents an item; Establishing a time period matrix corresponding to the multidimensional interest matrix; Calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix; The fusion weight is calculated based on the time period interest weight corresponding to the time period matrix and the real-time interest weight, wherein similarity is calculated based on the aggregation results of items in different time periods p and the multidimensional interest matrix, and the similarity is normalized as the historical interest vector weight of time period p; Calculate the real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix; Determine a target recommended item based on the real-time interest vector and the representation vector of the candidate item; The step of establishing a time period matrix corresponding to the multi-dimensional interest matrix includes: Obtaining the time corresponding to the historical viewing record, and determining the period of the time; Divide the cycle into a number of time periods, and aggregate the items corresponding to the historical viewing records in the same time period to obtain the historical viewing records corresponding to different time periods; A time period matrix corresponding to the multidimensional interest matrix is ​​established based on the historical viewing records of the different time periods and the multidimensional interest matrix.

2. The item recommendation method according to claim 1, characterized in that: The step of determining the multi-dimensional interest matrix of the family user based on the historical viewing records of the family user and determining the representation vector of each item includes: Acquire historical viewing records of the family user and feature data corresponding to the historical viewing records; generating a multi-dimensional interest matrix of the family user based on the historical viewing records; A representation vector corresponding to the item is generated based on the feature data corresponding to the historical viewing record.

3. The item recommendation method according to claim 1, characterized in that: The calculating the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix includes: Acquire the real-time playback information of the family user, and calculate the representation vector of the object in the real-time playback information; Calculating the real-time interest representation vector of the family user based on the representation vector of the item in the real-time broadcast information; The real-time interest weight of the family user is calculated according to the real-time interest representation vector of the family user and the multi-dimensional interest matrix.

4. The item recommendation method according to claim 1, characterized in that: The calculating the fusion weight based on the time period interest weight corresponding to the time period matrix and the real-time interest weight includes: Set adjustment parameters; The fusion weight is calculated based on the adjustment parameter, the time period interest weight corresponding to the time period matrix, and the real-time interest weight; The adjustment parameter is used to indicate the ratio of the real-time interest to the historical interest of the family user.

5. The item recommendation method according to claim 1, characterized in that: The determining of the target recommended item based on the real-time interest vector and the representation vector of the candidate item includes: The candidate items corresponding to the similarity exceeding the similarity threshold are selected as the target recommended items.

6. An item recommendation device, characterized in that: include: A first unit is used to determine a multidimensional interest matrix of the family user based on the historical viewing records of the family user, and determine a representation vector of each item, wherein the multidimensional interest matrix is ​​used to represent the viewing behavior of the user, and the multidimensional interest matrix is ​​a matrix composed of n m-dimensional vectors, wherein n is the number of interests of the family user, m is the vector dimension of the item, and each row of the multidimensional interest matrix represents an item; The second unit is used to establish a time period matrix corresponding to the multi-dimensional interest matrix; A third unit is used to calculate the real-time interest weight of the family user according to the item representation vector corresponding to the real-time playback information of the family user and the multi-dimensional interest matrix; The fourth unit is used to calculate the fusion weight based on the time period interest weight corresponding to the time period matrix and the real-time interest weight, wherein similarity is calculated based on the aggregation results of items in different time periods p and the multidimensional interest matrix, and the similarity is normalized as the historical interest vector weight of time period p; A fifth unit, configured to calculate a real-time interest vector of the family user based on the fusion weight and the multi-dimensional interest matrix; A sixth unit, configured to determine a target recommended item based on the real-time interest vector and a representation vector of the candidate item; The second unit is specifically used for: Obtaining the time corresponding to the historical viewing record, and determining the period of the time; Divide the cycle into a number of time periods, and aggregate the items corresponding to the historical viewing records in the same time period to obtain the historical viewing records corresponding to different time periods; A time period matrix corresponding to the multidimensional interest matrix is ​​established based on the historical viewing records of the different time periods and the multidimensional interest matrix.

7. An electronic device, characterized in that: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the item recommendation method as described in any one of claims 1 to 5.

8. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the item recommendation method as described in any one of claims 1 to 5 are implemented.

Images