阅读说明：本技术 一种烹饪过程中的意图识别方法和智能烹饪设备 (Intention identification method in cooking process and intelligent cooking equipment ) 是由 朱泽春 苗忠良 王�忠 于 2018-07-27 设计创作，主要内容包括：本申请提出一种烹饪过程中的意图识别方法和智能烹饪设备,所述方法包括：接收用户的语音信息并通过语义解析获取所述语音信息所对应的词典数据；获取烹饪设备传感器的状态参数并确定所述烹饪设备的工作状态数据；根据所述词典数据以及所述工作状态数据确定用户当前的意图。本发明通过对用户的语音信息进行语义解析,基于通过烹饪设备采集的状态参数确定的工作状态,判断用户的意图；提高烹饪设备的问答效率。(The application provides an intention identification method and intelligent cooking equipment in a cooking process, wherein the method comprises the following steps: receiving voice information of a user and obtaining dictionary data corresponding to the voice information through semantic analysis; acquiring state parameters of a sensor of a cooking device and determining working state data of the cooking device; and determining the current intention of the user according to the dictionary data and the working state data. According to the invention, the voice information of the user is subjected to semantic analysis, and the intention of the user is judged based on the working state determined by the state parameters collected by the cooking equipment; the question answering efficiency of the cooking equipment is improved.)

1. An intention recognition method in a cooking process, comprising:

receiving voice information of a user and obtaining dictionary data corresponding to the voice information through semantic analysis;

acquiring state parameters of a sensor of a cooking device and determining working state data of the cooking device;

and determining the current intention of the user according to the dictionary data and the working state data.

2. The method of claim 1, wherein: determining the current intent of the user from the dictionary data and the work state data comprises:

and determining the support degree S and the confidence degree C of the dictionary data and the working state data by a classification algorithm based on an association rule, and determining the current intention of the user according to the determined support degree S and the confidence degree C.

3. The method of claim 2, wherein determining the user's current intent from the dictionary data and the work state data comprises:

discretizing and missing value processing are carried out on the dictionary data and the working state data, and the support degree S and the confidence degree C between the dictionary data and the working state data are determined;

determining the current intention of the user according to the support degree S and the confidence degree C;

the dictionary data comprises dictionary attribution information, keyword slot position information and keyword semantic information which are needed for analyzing the voice information; the working state data comprises equipment type information, cooking progress information and cooking parameter information of the cooking equipment; the intentions include cooking related intentions related to cooking guidance or device control related to the cooking process, and dialogue quiz intentions related to encyclopedia knowledge not related to the cooking process.

4. The method of claim 3, wherein: the step of determining a support degree S and a confidence degree C between the dictionary data and the working state data includes:

calculating a support degree S and a confidence degree C between the dictionary data and the item set X and the working state data as an item set Y, wherein,

s (X → Y) ═ δ (X ∪ Y)/N, C (X → Y) ═ δ (X ∪ Y)/δ (X), N is the total number of terms in the X term set and the Y term set, δ (X ∪ Y) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is the same as the intention pointed to by the operating state data, and δ (X) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is.

5. The method of claim 4, wherein: the step of determining the current intention of the user according to the support degree S and the confidence degree C comprises the following steps:

when S is larger than or equal to Smin and C is larger than or equal to Cmin, judging that the current intention of the user is a cooking related intention;

when S is less than Smin and C is less than Cmin, judging that the current intention of the user is a question-answer intention;

when S is larger than or equal to Smin and C is smaller than Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

when S is less than Smin and C is more than or equal to Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

where Smin is the support threshold and Cmin is the confidence threshold.

6. The method of claim 1 or 2, wherein: further comprising:

if the current intention of the user is determined to be the existing intention of the cooking equipment, performing service response according to the identification of the existing intention;

and if the current intention of the user is determined to be the intention which does not exist in the cooking equipment, generating a new user intention identifier, and carrying out service response according to the new user intention identifier.

7. The method of claim 2, wherein: further comprising: when a plurality of cooking devices exist, respectively obtaining the support degree S and the confidence degree C of the dictionary data and the working state data of each cooking device;

determining the work flow of the cooking equipment with the maximum support degree S or confidence degree C as the optimal work flow;

and performing service response according to the optimal workflow.

8. The method of claim 1, wherein: further comprising:

acquiring user state parameters acquired by a robot sensor associated with the cooking equipment;

and determining the working mode of the cooking equipment according to the user state parameter.

9. An intelligent cooking device, comprising:

the semantic analysis module is used for receiving the voice information of the user and acquiring dictionary data corresponding to the voice information through semantic analysis;

the cooking system comprises a state module, a state detection module and a control module, wherein the state module is used for acquiring state parameters of a sensor of the cooking equipment and determining working state data of the cooking equipment;

an intent module configured to determine a current intent of the user based on the dictionary data and the operating state data.

10. The cooking apparatus of claim 9, wherein: the intention module determines the current intention of the user according to the dictionary data and the working state data, and comprises the following steps:

discretizing and missing value processing are carried out on the dictionary data and the working state data, and the support degree S and the confidence degree C between the dictionary data and the working state data are determined;

determining the current intention of the user according to the support degree S and the confidence degree C;

the dictionary data comprises dictionary attribution information, keyword slot position information and keyword semantic information which are needed for analyzing the voice information; the working state data comprises equipment type information, cooking progress information and cooking parameter information of the cooking equipment; the intentions include cooking related intentions related to cooking guidance or device control related to the cooking process, and dialogue quiz intentions related to encyclopedia knowledge not related to the cooking process.

Technical Field

The invention relates to the field of intelligent equipment control, in particular to an intention identification method in a cooking process and intelligent cooking equipment.

Background

In the cooking process of a user, the robot performs cooking guidance according to a preset workflow, at this time, if the user asks a question for the robot, the existing robot does not know whether the corresponding question is related to the current cooking scene due to lack of an intention judgment mechanism, and therefore, whether to provide content related to the current cooking scene or provide encyclopedic knowledge cannot be accurately judged, under the circumstance, the user may need to be subjected to multiple rounds of inquiry to supplement slot data, so that the cooking process is blocked, and the question-answering efficiency is low.

The defects of the current intelligent question-answering technical scheme are mainly reflected in that: the intention processing is still serial, and the support for parallel processing of a plurality of intention questions and answers of the same user is lacked; device context scenarios cannot be supported; the number of slots for multiple rounds of question answering is fixed, and the intention node processing and intention state tracking cannot be flexibly carried out according to equipment data, so that difficulties can be met when time-consuming scenes with definite time sequence such as cooking guidance are processed.

Disclosure of Invention

The invention provides an intention recognition method in a cooking process and intelligent cooking equipment, and aims to recognize the intention of voice information of a user in the cooking process.

In order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows:

in a first aspect, the present invention provides an intention recognition method in a cooking process, comprising:

receiving voice information of a user and obtaining dictionary data corresponding to the voice information through semantic analysis;

acquiring state parameters of a sensor of a cooking device and determining working state data of the cooking device;

and determining the current intention of the user according to the dictionary data and the working state data.

Preferably, determining the current intent of the user from the dictionary data and the work state data comprises:

and determining the support degree S and the confidence degree C of the dictionary data and the working state data by a classification algorithm based on an association rule, and determining the current intention of the user according to the determined support degree S and the confidence degree C.

Preferably, determining the current intent of the user from the dictionary data and the work state data comprises:

discretizing and missing value processing are carried out on the dictionary data and the working state data, and the support degree S and the confidence degree C between the dictionary data and the working state data are determined;

determining the current intention of the user according to the support degree S and the confidence degree C;

the dictionary data comprises dictionary attribution information, keyword slot position information and keyword semantic information which are needed for analyzing the voice information; the working state data comprises equipment type information, cooking progress information and cooking parameter information of the cooking equipment; the intentions include cooking related intentions related to cooking guidance or device control related to the cooking process, and dialogue quiz intentions related to encyclopedia knowledge not related to the cooking process.

Preferably, the step of determining a support degree S and a confidence degree C between the dictionary data and the working state data includes:

calculating a support degree S and a confidence degree C between the dictionary data and the item set X and the working state data as an item set Y, wherein,

s (X → Y) ═ δ (X ∪ Y)/N, C (X → Y) ═ δ (X ∪ Y)/δ (X), N is the total number of terms in the X term set and the Y term set, δ (X ∪ Y) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is the same as the intention pointed to by the operating state data, and δ (X) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is.

Preferably, the step of determining the current intention of the user according to the support degree S and the confidence degree C includes:

when S is larger than or equal to Smin and C is larger than or equal to Cmin, judging that the current intention of the user is a cooking related intention;

when S is less than Smin and C is less than Cmin, judging that the current intention of the user is a question-answer intention;

when S is larger than or equal to Smin and C is smaller than Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

when S is less than Smin and C is more than or equal to Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

where Smin is the support threshold and Cmin is the confidence threshold.

Preferably, the method further comprises:

if the current intention of the user is determined to be the existing intention of the cooking equipment, performing service response according to the identification of the existing intention;

and if the current intention of the user is determined to be the intention which does not exist in the cooking equipment, generating a new user intention identifier, and carrying out service response according to the new user intention identifier.

Preferably, the method further comprises: when a plurality of cooking devices exist, respectively obtaining the support degree S and the confidence degree C of the dictionary data and the working state data of each cooking device;

determining the work flow of the cooking equipment with the maximum support degree S or confidence degree C as the optimal work flow;

and performing service response according to the optimal workflow.

Preferably, the method further comprises:

acquiring user state parameters acquired by a robot sensor associated with the cooking equipment;

and determining the working mode of the cooking equipment according to the user state parameter.

In a second aspect, the present invention provides an intelligent cooking apparatus, comprising:

the semantic analysis module is used for receiving the voice information of the user and acquiring dictionary data corresponding to the voice information through semantic analysis;

the cooking system comprises a state module, a state detection module and a control module, wherein the state module is used for acquiring state parameters of a sensor of the cooking equipment and determining working state data of the cooking equipment;

an intent module configured to determine a current intent of the user based on the dictionary data and the operating state data.

Preferably, the intention module determines the current intention of the user according to the dictionary data and the working state data by:

discretizing and missing value processing are carried out on the dictionary data and the working state data, and the support degree S and the confidence degree C between the dictionary data and the working state data are determined;

determining the current intention of the user according to the support degree S and the confidence degree C;

the dictionary data comprises dictionary attribution information, keyword slot position information and keyword semantic information which are needed for analyzing the voice information; the working state data comprises equipment type information, cooking progress information and cooking parameter information of the cooking equipment; the intentions include cooking related intentions related to cooking guidance or device control related to the cooking process, and dialogue quiz intentions related to encyclopedia knowledge not related to the cooking process.

According to the invention, the voice information of the user is subjected to semantic analysis, and the intention of the user is judged based on the working state determined by the state parameters collected by the cooking equipment; the question answering efficiency of the cooking equipment is improved. Has the following beneficial effects:

1. after semantic analysis of voice information is carried out, recognition of user intention is carried out to match intention and response through collection and analysis of sensor data of cooking equipment;

2. the invention supports the scene of multi-intention parallel processing under the context environment of the cooking equipment in the cooking process of the user, and can switch among a plurality of intentions;

3. the support degree S and the confidence degree C of the dictionary data and the working state data are determined by a classification algorithm based on an association rule, the current intention of a user is determined according to the determined support degree S and the determined confidence degree C, and the method is very important and has important significance in an interactive scene of making intelligent question answering according to equipment context;

4. the invention supports discontinuous multi-round voice interaction in cooking guidance and other environments.

Drawings

Fig. 1 is a flowchart of an intention recognition method in a cooking process according to an embodiment of the present invention;

fig. 2 is a schematic structural diagram of an intelligent cooking device according to an embodiment of the invention;

FIG. 3 is a flow chart of intent recognition during cooking in accordance with an embodiment of the present invention;

fig. 4 is a schematic diagram of a cooking device and a cloud server system according to an embodiment of the present invention;

fig. 5 is a flowchart of the intention recognition performed by the cloud server according to the embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the following description of the embodiments of the present invention with reference to the accompanying drawings is provided, and it should be noted that, in the case of conflict, features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other.

As shown in fig. 1, an embodiment of the present invention provides an intention identification method in a cooking process, including:

s101, receiving voice information of a user and obtaining dictionary data corresponding to the voice information through semantic analysis;

s102, acquiring state parameters of a sensor of the cooking equipment and determining working state data of the cooking equipment;

s103, determining the current intention of the user according to the dictionary data and the working state data.

In the embodiment of the present invention, the order of steps S101 and S102 may be changed, and the manner of acquiring the status parameters of the sensor of the cooking apparatus in step S102 may be timed or triggered, for example, the status parameters may be acquired by the sensor every 30S, or the status parameters may be acquired by the sensor when the voice information of the user is received.

According to the invention, the voice information of the user is subjected to semantic analysis, and the intention of the user is judged based on the working state determined by the state parameters collected by the cooking equipment; the question answering efficiency of the cooking equipment is improved.

In the embodiment of the present invention, the calculation and the identification in steps S101 to S103 may be performed by a cooking device, or performed by a robot associated with the cooking device, or performed by a cloud server. The cooking equipment acquires voice information of a user through a microphone or a microphone array, and semantic analysis is performed by the cooking equipment or an associated robot or a cloud server to acquire corresponding dictionary data. The state parameters are analyzed by the cooking equipment or the associated robot or the cloud server to determine the working state, and the cooking equipment or the associated robot or the cloud server performs intention identification.

After step S101 receives the voice information of the user, the embodiment of the present invention further includes:

and matching the voice information with pre-stored control instruction information, and determining whether to perform intention identification according to a matching result.

If the voice instruction is matched with pre-stored control instruction information, skipping the step of intention identification and controlling the cooking equipment to execute the content of the voice information; if the voice command does not match the pre-stored control command information, a step of intention recognition is performed (step S103).

The semantic parsing process in step S101 may be:

converting voice information input by a user into a voice text;

and performing semantic understanding and semantic analysis on the voice text to obtain corresponding dictionary data.

In the embodiment of the present invention, the step S103 of determining the current intention of the user according to the dictionary data and the working state data includes:

and determining the support degree S and the confidence degree C of the dictionary data and the working state data by a classification algorithm based on an association rule, and determining the current intention of the user according to the determined support degree S and the confidence degree C.

In the embodiment of the present invention, a cba (classification base of association) algorithm based on a classification algorithm of an association rule is preferably used in a manner of determining dictionary data and working state data.

Step S103 of determining the current intention of the user according to the dictionary data and the operating state data includes:

discretizing and missing value processing are carried out on the dictionary data and the working state data, and the support degree S and the confidence degree C between the dictionary data and the working state data are determined;

determining the current intention of the user according to the support degree S and the confidence degree C;

the dictionary data comprises dictionary attribution information, keyword slot position information and keyword semantic information which are needed for analyzing the voice information; the working state data comprises equipment type information, cooking progress information and cooking parameter information of the cooking equipment; the intentions include cooking related intentions related to cooking guidance or device control related to the cooking process, and dialogue quiz intentions related to encyclopedia knowledge not related to the cooking process.

The embodiment of the invention divides words into a plurality of dictionary information lists according to categories in advance, such as a cooking information list, a control information list, a weather information list and the like, and determines which dictionary information list belongs to according to the content of voice information after receiving the voice information of a user, namely determines the attribution information of the dictionary. Then, the content of the voice information is divided into words and grammar components, the central component of the content of the voice information is determined according to the information of the main words, the predicate, the object, the shape, the complement and the like, namely the slot position information of the key words is determined, and finally the content of the voice information is subjected to semantic analysis according to the attributive dictionary information list to determine the semantic information of the key words.

For example, when the user inputs the voice information of "opening the braised pork recipe", the dictionary attribution information is determined as a "cooking information list", then word segmentation and grammar component segmentation are carried out, the keyword slot position information of "opening", "braised pork" and "recipe" is determined, and then the keyword semantic information is determined in the corresponding "cooking information list".

Specifically, the step of determining the support degree S and the confidence degree C between the dictionary data and the operating state data includes:

calculating a support degree S and a confidence degree C between the dictionary data and the item set X and the working state data as an item set Y, wherein,

s (X → Y) ═ δ (X ∪ Y)/N, C (X → Y) ═ δ (X ∪ Y)/δ (X), N is the total number of terms in the X term set and the Y term set, δ (X ∪ Y) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is the same as the intention pointed to by the operating state data, and δ (X) is the number of terms in which the intention pointed to by the statistically obtained dictionary data is.

In the embodiment of the present invention, the step of determining the current intention of the user according to the support degree S and the confidence degree C includes:

when S is larger than or equal to Smin and C is larger than or equal to Cmin, judging that the current intention of the user is a cooking related intention;

when S is less than Smin and C is less than Cmin, judging that the current intention of the user is a question-answer intention;

when S is larger than or equal to Smin and C is smaller than Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

when S is less than Smin and C is more than or equal to Cmin, starting a plurality of rounds of conversations to inquire so as to supplement corresponding keyword slot position information;

where Smin is the support threshold and Cmin is the confidence threshold.

In the embodiment of the present invention, the method further includes:

if the current intention of the user is determined to be the existing intention of the cooking equipment, performing service response according to the identification of the existing intention;

and if the current intention of the user is determined to be the intention which does not exist in the cooking equipment, generating a new user intention identifier, and carrying out service response according to the new user intention identifier.

When a plurality of cooking devices exist, respectively obtaining the support degree S and the confidence degree C of the dictionary data and the working state data of each cooking device;

determining the work flow of the cooking equipment with the maximum support degree S or confidence degree C as the optimal work flow;

and performing service response according to the optimal workflow.

The method of the embodiment of the invention also comprises the following steps:

acquiring user state parameters acquired by a robot sensor associated with the cooking equipment;

and determining the working mode of the cooking equipment according to the user state parameter.

With the development of current cooking equipment and robots, the cooking equipment can realize more functions, for example, a range hood can control the rotating speed of a fan, and can also control the brightness of light, or communicate with wearable equipment. Or the wearable device is used for detecting the physical sign data (heart rate, blood pressure, electrocardiogram) of the user, determining the state of the user according to the physical sign data, and further determining the working mode of the cooking device, for example, the brightness of the light can be adjusted or music can be played.