Complex sentence query method, device and medium applied to in-vehicle scenarios

Abstract

The application relates to a complex sentence query method, equipment and medium applied to an in-vehicle scene. The complex sentence query method comprises the following steps: after it is determined that a complex sentence is received, a plurality of target keywords in the complex sentence are parsed; fields to which the target keywords respectively belong are queried to obtain at least one field; the query relationship between the target keywords under the same field is set as OR, and the plurality of target keywords are sequentially queried according to the priority between the at least one field to output at least one query result. By adopting the complex sentence query method, the complex sentence query in the in-vehicle scene can be guaranteed to have a result output, and the real-time requirement of the in-vehicle scene can be met.

Description

Technical Field

[0001] This application relates to the field of automotive technology, and in particular to a method, device, and medium for querying complex statements in an in-vehicle setting. Background Technology

[0002] While driving, drivers can speak queries as needed to obtain the required results. Due to the high speed, queries in the in-vehicle environment need to output relatively accurate results in a short time; otherwise, drivers may miss the correct exit, or their attention may be severely distracted due to repeated attempts or prolonged searching without finding the desired result, thus affecting safe driving.

[0003] Currently, when querying complex user queries, there are instances where no results are returned. This is because complex queries often involve numerous and semantically intricate conditions. While recommendation algorithms could be used to address this, their complexity makes them unsuitable for the real-time requirements of in-vehicle queries.

[0004] It is evident that a method for querying complex statements in in-vehicle scenarios is currently lacking. Summary of the Invention

[0005] Based on this, this application provides a method, device and medium for complex statement querying in in-vehicle scenarios, which can ensure that complex statement queries in in-vehicle scenarios can output results and also meet the real-time requirements of in-vehicle scenarios.

[0006] Firstly, this application provides a method for querying complex statements in an in-vehicle scenario. The method includes: after determining that a complex statement has been received, parsing out multiple target keywords in the complex statement; querying the fields to which each target keyword belongs to obtain at least one field; setting the query relationship between target keywords under the same field as OR, and querying multiple target keywords step by step according to the priority between at least one field to output at least one query result.

[0007] In conjunction with the first aspect, in the first possible implementation of the first aspect, the aforementioned step-by-step query of multiple target keywords according to the priority between at least one field to output at least one query result includes: performing step-by-step query of target keywords under each field in at least one field in descending order of priority of at least one field, and monitoring the returned query results; when a query result is detected, outputting the query result returned by the current level query, and stopping the query of target keywords in the next level field of at least one field.

[0008] In conjunction with the first aspect, in the second possible implementation of the first aspect, the aforementioned step-by-step query of multiple target keywords according to the priority of at least one field to output at least one query result includes: performing step-by-step query of target keywords under each field in at least one field in descending order of priority of at least one field, and monitoring the returned query results; when a query result is detected, repeating the step of querying the target keywords of the next level field in at least one field based on the query result of the current level query, until no search results are returned, and then outputting the query result returned by the previous level query.

[0009] In conjunction with the first aspect, in the third possible implementation of the first aspect, before performing a step-by-step query on multiple target keywords according to the priority between at least one field, the method further includes: querying the initial priority corresponding to each of the at least one field, and determining whether there are multiple target fields among the at least one field, wherein the initial priorities of the multiple target fields are equal; if so, determining the relative priority among the multiple target fields according to user preferences; and determining the priority among the at least one field according to the initial priority corresponding to each of the at least one field and the relative priority among the multiple target fields.

[0010] In conjunction with the third possible implementation of the first aspect, in the fourth possible implementation of the first aspect, the aforementioned determination of the relative priority among multiple target fields based on user preferences includes: calculating the click component of each target field and the total click volume of peer fields, wherein peer fields are all fields in the corpus that have the same initial priority as the multiple target fields; calculating the ratio of the click component to the total click volume of each target field to obtain the click percentage of each target field, and comparing the numerical relationship between the click percentages of each target field; and determining the relative priority among the multiple target fields according to the numerical relationship.

[0011] In conjunction with the fourth possible implementation of the first aspect, in the fifth possible implementation of the first aspect, when there are target fields with the same click percentage among multiple target fields, the method further includes: determining the order of target fields with the same click percentage according to the position order of target keywords under each target field in the complex statement; and determining the relative priority of target fields with the same click percentage according to the order.

[0012] In conjunction with the first aspect, in the sixth possible implementation of the first aspect, before determining that a complex statement has been received, the method further includes: after receiving the user's query statement, querying the query statement; if no query results are returned, detecting the number of target keywords in the query statement; if the number of target keywords in the query statement is detected to be greater than or equal to two, determining that the query statement is a complex statement, and determining that a complex statement has been received.

[0013] Secondly, this application also provides a query device, which includes: a parsing unit, configured to parse out multiple target keywords in a complex statement after determining that a complex statement has been received; a determining unit, configured to query the fields to which each target keyword belongs to obtain at least one field; and a querying unit, configured to set the query relationship between target keywords under the same field to OR, and to perform a step-by-step query on the multiple target keywords according to the priority between the aforementioned at least one field, so as to output at least one query result.

[0014] In conjunction with the second aspect, in the first possible implementation of the second aspect, the aforementioned query unit is specifically used to: perform a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of at least one field, and monitor the returned query results; when a query result is detected, output the query result returned by the current level query, and stop querying the target keywords of the next level field in the at least one field.

[0015] In conjunction with the second aspect, in the second possible implementation of the second aspect, the aforementioned query unit is specifically used to: perform a step-by-step query on the target keywords under each field in the at least one field according to the priority of at least one field from high to low, and monitor the returned query results; when a query result is detected, repeat the step of querying the target keywords of the next level field in the at least one field based on the query result of the current level query, until no search results are returned, and output the query result returned by the previous level query.

[0016] In conjunction with the second aspect, in a third possible implementation of the second aspect, the aforementioned determining unit is further configured to: query the initial priority corresponding to at least one field, and determine whether there are multiple target fields among the at least one field, wherein the initial priorities of the multiple target fields are equal; if so, determine the relative priority among the multiple target fields according to user preferences; and determine the priority among at least one field according to the initial priority corresponding to at least one field and the relative priority among the multiple target fields.

[0017] In conjunction with the third possible implementation of the second aspect, in the fourth possible implementation of the second aspect, the aforementioned determining unit is specifically used for: counting the click component of each target field and the total click volume of peer fields, wherein peer fields are all fields in the corpus that have the same initial priority as the multiple target fields; calculating the ratio of the click component of each target field to the total click volume to obtain the click percentage of each target field, and comparing the numerical relationship between the click percentages of each target field; and determining the relative priority among the multiple target fields according to the numerical relationship.

[0018] In conjunction with the fourth possible implementation of the second aspect, in the fifth possible implementation of the second aspect, when there are target fields with the same click percentage among multiple target fields, the aforementioned determining unit is further configured to: determine the order of target fields with the same click percentage according to the position order of the target keywords under each target field in the complex statement; and determine the relative priority of target fields with the same click percentage according to the order.

[0019] In conjunction with the second aspect, in the sixth possible implementation of the second aspect, before determining that a complex statement has been received, the determining unit is further configured to: after receiving the user's query statement, perform a query on the query statement; if no query result is returned, detect the number of target keywords in the query statement; if the number of target keywords in the query statement is detected to be greater than or equal to two, determine that the query statement is a complex statement, and determine that a complex statement has been received.

[0020] Thirdly, this application also provides a query device, which includes a processor, a transceiver, and a memory, the processor and the memory being connected via a bus; the processor is used to execute multiple instructions; the transceiver is used to interact with other devices; the memory is used to store multiple instructions, the instructions being adapted to be loaded by the processor and executed as a complex statement query method applied to an in-vehicle scenario, such as the first aspect or any embodiment of the first aspect.

[0021] Fourthly, this application also provides a computer-readable storage medium storing multiple instructions adapted for loading and execution by a processor of a complex statement query method applied in a vehicle-in-the-vehicle scenario, as described in the first aspect or any embodiment of the first aspect.

[0022] In summary, this application provides a method, device, and medium for complex statement queries in in-vehicle scenarios. The complex statement query method employs a degenerate query approach, converting complex statements into simpler ones to ensure at least one query result is output within a short time. Specifically, after receiving a complex query statement, the query device first extracts multiple target keywords from the complex statement, sets the query relationship between target keywords with the same field to "OR," and then performs a step-by-step query on the target keywords according to the priority of each field to output at least one query result. Therefore, by implementing the method of this application, not only is at least one query result output guaranteed, but also, due to its simplicity, the query result can be output quickly. In conclusion, this application provides a complex query method for in-vehicle scenarios, ensuring not only the output of results for complex statement queries in in-vehicle scenarios but also meeting the real-time requirements of in-vehicle environments. Attached Figure Description

[0023] Figure 1 This is a schematic flowchart illustrating a complex statement query method in one embodiment of this application;

[0024] Figure 2 This is a schematic flowchart illustrating the process of determining the priority of a field in one embodiment of this application;

[0025] Figure 3 This is a schematic block diagram of a query device in one embodiment of this application;

[0026] Figure 4 This is a structural block diagram of a query device in one embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0028] To address the issue that current query methods either fail to return results or do not meet the real-time requirements of in-vehicle scenarios when querying complex statements, this application proposes a complex statement query method for in-vehicle scenarios that can improve upon these problems. Specifically, the complex statement query method provided in this application includes: after confirming that a complex statement has been received, parsing out multiple target keywords from the complex statement; querying the fields to which each target keyword belongs to obtain at least one field; setting the query relationship between target keywords under the same field as OR; and performing a step-by-step query on multiple target keywords according to the priority between at least one field to output at least one query result.

[0029] It should be noted that the core of the complex statement query method in this application lies in the degenerate query performed on the complex statement. When performing the degenerate query, the query device performs a step-by-step query on multiple target keywords of the complex statement according to the priority of each field. During this step-by-step query, the query relationship between target keywords under the same field is set to "OR". For example, target keyword A and target keyword B in the complex statement both belong to field C. Therefore, when the query device reaches field C, it queries target keywords A and B under field C using the query formula "A OR B". It is evident that because the implementation of this application is simple and at least one query result is output, implementing the complex statement query method of this application not only ensures that complex statement queries in in-vehicle scenarios have result output but also meets the real-time requirements of in-vehicle scenarios. Furthermore, the "multiple" described in this application refers to two or more.

[0030] It should also be noted that the complex statement query method provided in this application not only ensures query results but also balances accuracy and recall. Specifically: on the one hand, because this application queries multiple target keywords of complex statements according to the priority between fields, rather than using an "OR" query method, the returned query results are more accurate, thus improving accuracy. On the other hand, because this application sets the query relationship between target keywords under the same field to "OR," instead of using a priority-based query method, more correct query results can be returned, thus improving recall. Accuracy refers to the ratio of correctly returned query results to all returned query results, and recall refers to the ratio of correctly returned query results to all correct query results.

[0031] To better understand the complex statement query method proposed in this application for in-vehicle scenarios, such as Figure 1 As shown, this application proposes one embodiment of the method. Next, it will be discussed in conjunction with... Figure 1 The flowchart shown illustrates the complex statement query method proposed in this application. Specifically:

[0032] 101: After confirming that a complex statement has been received, parse out multiple target keywords in the complex statement.

[0033] The complex statement described in this application can be a query statement that includes two or more target keywords that match the keywords in the corpus. After the query device determines that it has received the complex statement, it matches the keywords in the corpus with the complex statement to parse out the target keywords included in the complex statement. For example, suppose a user sends the complex statement "I want to watch 'Film X' starring actor A and actor B". After matching the complex statement with the keywords in the corpus, the three target keywords "actor A", "actor B" and "Film X" can be determined.

[0034] It should be noted that, in order to quickly identify complex statements, before confirming that a complex statement has been received, the received user query statement can first be queried using conventional query methods. If no query results are returned, the complexity of the query statement is further determined by judging whether it contains two or more target keywords. If multiple keywords are included, a complex statement is confirmed to have been received, and then the complex statement query of this application is triggered, that is, multiple keywords in the complex statement are parsed as target keywords. Specifically, in one implementable method, before confirming that a complex statement has been received, the query device queries the user query statement after receiving it; if no query results are returned, the number of target keywords in the query statement is detected; if the number of target keywords in the query statement is greater than or equal to two, the query statement is confirmed to be a complex statement, and it is confirmed that a complex statement has been received.

[0035] It should also be noted that the corpus provided in this application may include multiple query domains, each of which includes a large number of pre-stored fields and a large number of keywords under each field. Accordingly, after determining that a complex statement has been received, the query device can use semantic analysis to select the query domain to be entered, and use all the keywords in the query domain to parse out multiple target keywords from the complex statement. Specifically, the query domains provided in this application include: "vehicle settings," "vehicle information," "multimedia," "map navigation," "electronic manual," and "calendar," etc. Table 1 provides examples of complex statements under different query domains.

[0036] Query domain Complex query examples Vehicle settings Open the camping mode and charging settings management page Vehicle Information I want to see the information on my car's refrigerant and brake fluid levels. multimedia I want to watch "Film X," a movie starring actor A and actor B. Map Navigation I'm looking for a cave hotel with hot springs. Electronic manual How do I change a spare tire on the highway? Schedule Calendar I'd like to see the schedule for August 1st.

[0037] Table 1

[0038] 102: Query the fields to which each target keyword belongs to obtain at least one field.

[0039] The complex statement contains multiple target keywords, which may belong to the same field or different fields. After parsing the multiple target keywords in the complex statement, the query device searches the corpus for the field to which each target keyword belongs to, so as to obtain at least one field.

[0040] It should be noted that the corpus also stores the initial priority of the fields. Therefore, the query device can also query the initial priority of any one of the aforementioned fields in the corpus. Then, the priority between the aforementioned at least one field can be determined based on the initial priority of each field, or the priority between the aforementioned at least one field can be determined based on user preferences (e.g., click percentage). This application does not limit this.

[0041] It should also be noted that the fields described in this application and their initial priorities are set according to different query domains within the in-vehicle scenario. The rationality of these settings is crucial to the accuracy of the query results. Therefore, this application not only subdivides various query domains within the automotive scenario but also sets the initial priorities of fields within each query domain according to multiple classification rules. These classification rules are non-user-related and include: classifying the initial priorities of fields based on their importance or functional level. Next, this application will use the "vehicle settings" domain and the "multimedia" domain as examples to explain in detail the field classification and initial priority settings for each query domain:

[0042] For example, in the "Vehicle Settings" field, the initial priority of each field is set according to its functional hierarchy, based on the functional division of vehicle settings. The fields in the vehicle settings field are assigned an initial priority from highest to lowest functional level, meaning that the outermost functional level field has the highest initial priority. This setting is to improve the accuracy of search results. Because outer functional interfaces often provide hints about inner functional interfaces, if the user is prioritized to return to the outer functional interface, they can use these hints to further access the inner functional interface and select the function they actually need. However, if the user is prioritized to return to the inner functional interface, they will not receive the hints and will not be able to find the function they need. Therefore, setting the initial priority from highest to lowest functional level improves the accuracy of search results.

[0043] Furthermore, the functional hierarchy of each field in the "Vehicle Settings" field includes three levels. Specifically: the first level of functional settings under the "Vehicle Settings" field includes vehicle operation settings, intelligent driving settings, sound settings, display settings, Bluetooth settings, language settings, mode settings, accessory settings, and general settings; taking vehicle operation settings as an example, the second level of functional settings under vehicle operation settings includes driving settings, lighting settings, door and window settings, vehicle lock settings, rearview mirror settings, steering wheel settings, seat settings, and maintenance settings; taking driving settings as an example, the third level of functional settings under driving settings includes convenient entry and exit settings, suspension comfort settings, steering mode settings, and energy recovery settings.

[0044] For example, in the "Multimedia" domain, the initial priority of fields in the "Multimedia" domain is set according to their importance. Each field in the Multimedia domain is assigned an initial priority from highest to lowest importance; that is, the more important the field, the higher its priority, in order to improve the accuracy of the query results.

[0045] Furthermore, the importance of each field in the "Multimedia" field is divided into three levels. Specifically, the first level under the "Multimedia" field includes video album name / alias, the second level includes director, actor, and role, the third level includes production company, payment type, recommended viewing age, rating, awards, series, personalized tags, rating, latest, high score / positive reviews / hottest / popular, and the fourth level includes channel, year, release time, region, and language.

[0046] 103: Set the query relationship between target keywords under the same field to OR, and perform a step-by-step query on the multiple target keywords according to the priority between at least one of the aforementioned fields, so as to output at least one query result.

[0047] In this process, after determining the fields to which the multiple target keywords belong, the query device performs a degenerate query on these target keywords. First, it sets the relationship between target keywords in the same field to "OR". Then, it queries the multiple target keywords in the complex statement level by level, according to the priority of at least one of the aforementioned fields from highest to lowest. This level-by-level query starts with the target keyword under the highest priority field among the aforementioned fields. If no result is found, it then queries the target keyword under the second highest priority field among the aforementioned fields. If still no result is found, it then queries the target keyword under the third highest priority field among the aforementioned fields, and so on, until a result is found. At this point, the query result returned by the current level can be directly output as a degenerate query, or the query can continue to the next level based on the result returned by the current level, and so on, until no result is output. Then, the query result of the previous level is output, thus completing multiple degenerate queries.

[0048] It should be noted that the complex query method used in this application is essentially a degenerate query, that is, it simplifies the query conditions of complex queries to balance accuracy and real-time performance. This application does not restrict whether a single or multiple degenerate queries are used when querying complex statements. The number of degenerate queries can be set according to the user's performance preferences; either a single or multiple degenerate queries can be used. Specifically: when the user prioritizes real-time performance, a single degenerate query is used; when the user prioritizes accuracy, multiple degenerate queries are used. The following section will explain the single and multiple degenerate query methods:

[0049] In one implementation, the query device performs a degenerate query on multiple target keywords of a complex statement. Specifically, the aforementioned step of performing a step-by-step query on multiple target keywords according to the priority of at least one field to output at least one query result includes: performing a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of the at least one field, and monitoring the returned query results; when a query result is detected, outputting the query result returned by the current level query, and stopping the query on the target keywords of the next level field in the at least one field.

[0050] In this method, the query device performs a step-by-step query on multiple target keywords in a complex statement according to the priority of at least one of the aforementioned fields. If no query result is returned at the current level, the next level query is executed... Once a query result is returned, a degenerate query is completed, and the currently returned query result is immediately output, without further querying of the remaining target keywords. This degenerate query in this embodiment can improve the emphasis on real-time performance while balancing accuracy and real-time performance.

[0051] For example, in the complex statement "I want to watch 'Film X' starring Actor A and B", "Film X" belongs to the "Film Title" field, while the target keywords "Actor A" and "Actor B" both belong to the "Actor" field. Therefore, the query relationship between "Actor A" and "Actor B" is set to "OR", that is, "Actor A or Actor B". Assuming that the priority of the "Actor" field is higher than that of the "Film Title" field, the query device first queries the target keyword "Actor A or Actor B" under the "Actor" field. If the query returns query results related to "Actor A" and / or "Actor B", a degenerate query is completed, and the next level of query is stopped. The query results related to "Actor A" and / or "Actor B" are directly output to complete the query of the complex statement "I want to watch 'Film X' starring Actor A and Actor B".

[0052] In another implementation, the query device performs multiple degenerate queries on multiple target keywords of a complex statement. Specifically, the aforementioned step of performing a step-by-step query on multiple target keywords according to the priority of at least one field to output at least one query result includes: performing a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of the at least one field, and monitoring the returned query results; when a query result is detected, repeating the step of querying the target keywords of the next level field in the at least one field based on the query result of the current level query, until no search results are returned, and then outputting the query result returned by the previous level query.

[0053] In this method, the query device sequentially queries multiple target keywords of a complex statement according to the priority of at least one of the aforementioned fields. If no query result is returned at a given level, the query proceeds to the target keywords under the next lower level field, and so on. Once a query result is returned, a degenerate query is completed. Based on the result of this degenerate query, the query continues to the target keywords under the next lower level field, and so on, until no query result is returned. Finally, the query result of the previous level field is output, which is the result of the previous degenerate query. This implementation of multiple degenerate queries can improve the emphasis on accuracy while balancing accuracy and real-time performance.

[0054] Following the example of the previous implementation, after a degenerate query, query results related to "Actor A" and / or "Actor B" are obtained. Based on these results, the query device continues to query the target keyword "Film / TV Series X" under the next-level field "Film / TV Series Name" to obtain query results related to "Film / TV Series X + Actor A", thus completing two degenerate queries. If there is another next-level field, the target keyword of the next-level field is queried based on the query results related to "Film / TV Series X + Actor A" to perform a third degenerate query. If no results are returned in the third degenerate query, the query results related to "Film / TV Series X + Actor A" are output to complete the query for the complex statement "I want to watch 'Film / TV Series X' starring Actor A and B".

[0055] Furthermore, when outputting query results, the results can be displayed in various ways. Taking the multimedia field as an example, in one implementation, the query device can adjust the priority of the output method based on vehicle information. Specifically: before outputting the query results, the query device can obtain current vehicle information through the Electronic Control Unit (ECU). This information includes the vehicle screens / vehicle status, with the driver's screen, passenger screen, and rear screens as examples. The vehicle status includes driving status and non-driving status. The priority of the output method is dynamically adjusted based on the current vehicle information. For example, if a user searches for "children's story version of Journey to the West," in driving status, the audio output method has a higher priority than the video output method, so an audio version of "Journey to the West" is output. In non-driving status, the video output method has a higher priority than the audio output method, so a video version of "Journey to the West" is output. For another example, when the vehicle has multiple screens, the audio output method of the driver's screen has a higher priority than the video output method, while the video output method of the passenger screen and rear screens has a higher priority than the audio output method, as detailed below:

[0056]

[0057] Table 2

[0058] In another possible implementation, when outputting the query results, the output method can be selected according to user habits. Specifically, before outputting the query results, the query device determines the user's habits based on the user's profile and adjusts the priority accordingly. For example, if the passenger searches for "children's version of Journey to the West" via voice, and the passenger's voiceprint shows a preference for or playback history of audio versions of Journey to the West, the audio will be played first.

[0059] In summary, this application embodiment adopts a degenerate query method when querying complex statements. The relationship between target keywords under the same field in the complex statement is set to "OR", and multiple target keywords in the complex statement are queried step by step according to the priority of each field to output at least one query result. This ensures that complex statement queries in the in-vehicle scenario have results output. Furthermore, because it is simple to implement, it can also meet the real-time requirements of the in-vehicle scenario.

[0060] To further improve the accuracy of complex query results in in-vehicle scenarios, based on the aforementioned embodiments, this application, before performing a step-by-step query on multiple target keywords according to the priority of at least one field, can also determine the priority between each field based on initial priority and relative priority. For example... Figure 2 As shown, this application also provides an embodiment of the process for determining the priority of fields. In this embodiment, the setting of field priorities is improved: on the one hand, an initial priority is pre-set for each field in the corpus according to the classification rules of each domain. It should be noted that there is not a one-to-one correspondence between fields and initial priorities; different fields may correspond to the same initial priority. On the other hand, a relative priority is also set for fields with the same initial priority according to user preferences, so that in the case of different initial priorities, the priority between fields can be determined according to the initial priority, while in the case of the same initial priority, the priority between fields can be determined according to the relative priority. Next, this embodiment will be combined with... Figure 2 To provide a detailed explanation, specifically:

[0061] 201: Query the fields to which each target keyword in a complex statement belongs, to obtain at least one field, and query the initial priority of each of the at least one field.

[0062] In this context, a target field refers to a field that has the same initial priority among at least one of the aforementioned fields. For example, suppose the complex statement "I want to watch 'Film X' starring Actor A and B" has an initial priority of L for the "Actor" field (belonging to "Actor A" and "Actor B"), and the initial priority of the "Film / TV Show Name" field (belonging to "Film / TV Show X") is also L. Then, both the "Actor" field and the "Film / TV Show Name" field are target fields. The relative priority among multiple target fields cannot be determined solely by their initial priority; therefore, the relative priority among these multiple target fields needs to be determined in subsequent steps.

[0063] 202: Determine whether there are multiple target fields among the aforementioned at least one field. If not, proceed to step 203; if so, proceed to step 204.

[0064] 203: If there are no multiple target fields among the aforementioned at least one field, the priority among the aforementioned at least one field shall be determined according to the initial priority corresponding to each field.

[0065] When the initial priorities of each of the aforementioned at least one field are not the same, the priority among the aforementioned at least one field can be determined based on the initial priority of each field.

[0066] For example, suppose in the complex statement "I want to watch 'Film X' starring Actor A and B", the initial priority of the "Actor" field, which contains "Actor A" and "Actor B", is L1, while the initial priority of the "Film Title" field, which contains "Film X", is L2. In this case, the initial priorities of the "Actor" field and the "Film Title" field are inconsistent. Therefore, the priority between the "Actor" field and the "Film Title" field can be determined based on the relationship between L1 and L2. That is, if L1 is greater than L2, the priority of the "Actor" field is set higher than the priority of the "Film Title" field; if L1 is less than L2, the priority of the "Film Title" field is set higher than the priority of the "Actor" field.

[0067] 204: In the case where there are multiple target fields in at least one of the aforementioned fields, determine the relative priority among the multiple target fields in the complex statement based on user preferences.

[0068] If multiple target fields exist among the aforementioned at least one field, the query device can determine the relative priority among these target fields based on user preference. For example, it can determine the relative priority based on user clicks on the target fields and / or the order of target keywords within the target fields in complex sentences. It's important to note that since users generally place the target keywords they care about most first when speaking, the order of target keywords within each target field in complex sentences also reflects the user's preference for each target field.

[0069] Next, to better understand the process of determining the relative priority proposed in this application, this application will describe it through three implementation methods, specifically:

[0070] In the first implementation, the query device adjusts the relative priority between fields by monitoring the number of user clicks on each field in real time. Specifically, the aforementioned step of determining the relative priority between multiple target fields based on user preferences includes: calculating the click component of each target field and the total number of clicks on peer fields, where peer fields are all fields in the corpus that have the same initial priority as the multiple target fields; calculating the ratio of the click component to the total number of clicks for each target field to obtain the click percentage of each target field, and comparing the numerical relationship between the click percentages of each target field; and determining the relative priority between multiple target fields according to the numerical relationship.

[0071] The query device can monitor the click volume of all keywords in the corpus in real time, and calculate the click component of each field based on the keyword click volume. Then, it sums the click components of all fields with the same initial priority to obtain the click component of the peer field. Specifically, after monitoring changes in the click volume of keywords in the corpus, the query device calculates the click component of each field, which is the sum of the click volumes of all keywords under each field in the corpus. Then, it calculates the total click volume of peer fields based on the click components of each field, which is the sum of the click components of all fields with the same initial priority. Finally, it calculates the ratio of the click component to the total click volume to obtain the click percentage of each field. Based on this click percentage, the relative priority among peer fields can be determined. It should be noted that peer fields refer to fields with the same initial priority. The keyword click volume refers to the number of times a user selects a keyword through clicking or voice, etc. The selection method is not limited to clicking; other methods of selecting keywords should be within the scope of protection of this application.

[0072] For example, as shown in Table 2 below, the corpus contains fields A, B, C, D, and E. Fields A, B, and C are of the same priority level (initial priority L1), while fields D and E are of the same priority level (initial priority L2). Monitoring shows that the click percentages of fields A, B, and C at the same priority level are 10%, 65%, and 25%, respectively. Therefore, the relative priorities of fields A, B, and C can be determined as L1*1.1, L1*1.65, and L1*1.25, respectively. The order of relative priority from highest to lowest is: C, B, and A. Similarly, the click percentages of fields D and E at the same priority level are 40% and 60%, respectively. Therefore, the relative priorities of fields D and E can be determined as L1*1.4 and L1*1.6, respectively. The order of relative priority from highest to lowest is: C, B, and A.

[0073] Field D.

[0074] Fields A B C D E total Initial priority L1 L1 L1 L2 L2 / Same-level click share 10％ 65％ 25％ 40％ 60％ 100％ relative priority L1*1.1 L1*1.65 L1*1.25 L2*1.4 L2*1.6 /

[0075] Table 3

[0076] In the second implementation, the query device determines the relative priority among multiple target fields based on the position order of the target keywords in the complex statement. Specifically, the aforementioned step of determining the relative priority among multiple target fields based on user preferences includes: determining the order of each target field according to the position order of the target keywords under each target field in the complex statement; and determining the relative priority among each target field according to the order.

[0077] The query device can directly determine the relative priority of each target field based on the position of the target keywords in the complex statement. For example, assuming the initial priority of the "actors" field and the "movie / film title" field is the same, when querying the complex statement "I want to watch 'Movie / Film X' starring actors A and B", it cannot determine whether to query "actors A" and "actors B" under the "actors" field first or "movie / film X" under the "movie / film title" field based on the initial priority. In this case, the query device will determine that the relative priority of the "actors" field is higher than the relative priority of the "movie / film title" field because "actors A" and "actors B" are located before "movie / film X" in the complex statement, and thus query "actors A" and "actors B" under the "actors" field first.

[0078] It should be noted that since some target fields may have multiple target keywords, and these keywords may not be adjacent but distributed in different positions within a complex statement, the position of the first target keyword can be used as the basis for sorting the target fields, thus determining the relative priority between them. For example, assuming "Actor A" precedes "Film / TV Series X" and "Actor B" follows "Film / TV Series X," then the priority of the "Actor" field is determined to be higher than that of the "Film / TV Series Name" field based on the order between "Actor A" and "Film / TV Series X."

[0079] The third implementation combines the two methods described above. The query device can first determine the relative priority based on the number of user clicks for each field. When the relative priority cannot be determined based on the number of user clicks, for example, when fields with the same initial priority have the same click percentage, the relative priority is then determined based on the position order of the target keywords in the complex sentence. Specifically, the aforementioned steps for determining the relative priority among multiple target fields based on user preferences include: calculating the click component of each target field and the total click volume of peer fields, where peer fields are all fields in the corpus with the same initial priority as the multiple target fields; calculating the ratio of the click component to the total click volume of each target field to obtain the click percentage of each target field, and comparing the numerical relationship between the click percentages of each target field; determining the relative priority among multiple target fields according to the numerical relationship; in the case where there are target fields with the same click percentage among multiple target fields, determining the order of the target fields with the same click percentage according to the position order of the target keywords under each target field in the complex sentence; and determining the relative priority among the target fields with the same click percentage according to the order.

[0080] For example, suppose the initial priority of the "Actor" field and the "Film / TV Show Name" field is both L, and the click share of the "Actor" field and the "Film / TV Show Name" field is both 10%. Therefore, it is impossible to determine the relative priority of the "Actor" field and the "Film / TV Show Name" field based on the initial priority and click share. In this case, we can determine that the "Actor" field comes before the "Film / TV Show Name" field based on the order of "Actor A" and "Actor B" under the "Actor" field and "Film / TV Show X" under the "Film / TV Show Name" field in the complex statement. Therefore, we set the relative priority of the "Actor" field to be greater than the relative priority of the "Film / TV Show Name" field.

[0081] 205: Determine the priority among the aforementioned at least one field based on the initial priority corresponding to each of the aforementioned at least one field and the relative priority among the aforementioned multiple target fields.

[0082] The query device first sorts the aforementioned at least one field according to its initial priority. If there are multiple target fields with the same initial priority among the at least one field, then the multiple target fields are sorted according to their relative priority to determine the priority among the at least one field.

[0083] Furthermore, when outputting query results, if there are multiple query results to be output, the number of output results can be determined according to user preferences. For example, the query device first determines the upper limit of the number of results to be displayed based on the size of the display screen, and then determines the number of query results related to each target keyword to be displayed based on the ratio of clicks between target keywords. This can improve the problem of information overload, minimize driver fatigue, and improve driving safety. For example, assuming the upper limit of the number of information that can be displayed on the screen is N (N≥2), there are M1 query results related to "actor A" and M2 query results related to "actor B", then M1×N / (M1+M2) query results related to "actor A" and M2×N / (M1+M2) query results related to "actor A" will be output on the screen. If the calculated result has a score, the result will be rounded up or down, with the score not exceeding N.

[0084] Furthermore, the query device can also determine the display order of query results corresponding to each target keyword based on the position and order of the target keywords in the complex statement. For example, in a complex statement, "actor A" comes before "actor B". Therefore, when outputting query results, the query results related to "actor A" will be displayed in the upper half of the screen, and the query results related to "actor B" will be displayed in the lower half of the screen.

[0085] In summary, to improve the accuracy of complex query queries, this application employs a two-tiered priority setting for fields. The priority of each field is determined by its initial priority and the relative priority between fields. It should be noted that the initial priority is a pre-set fixed value, while the relative priority is dynamically adjusted later. This combination of priorities ensures that the accuracy requirements for complex query queries in in-vehicle scenarios are basically met in the early stages when user feedback data is insufficient. Furthermore, accuracy can be further improved later based on a large amount of user feedback data. Therefore, this application's two-tiered priority setting effectively improves the accuracy of complex query queries in in-vehicle scenarios.

[0086] This application also provides a query device, such as Figure 3As shown. This application embodiment can divide the device into functional modules based on the above method example. For example, each function can be divided into its own functional module, or two or more functions can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods. Specifically, as... Figure 3 As shown, the query device includes: a parsing unit 310, used to parse out multiple target keywords in a complex statement after determining that a complex statement has been received; a determining unit 320, used to query the fields to which each target keyword belongs to obtain at least one field; and a querying unit 330, used to set the query relationship between target keywords under the same field to OR, and to perform a step-by-step query on the multiple target keywords according to the priority between the aforementioned at least one field, so as to output at least one query result.

[0087] In another possible implementation, the aforementioned query unit 330 is specifically used to: perform a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of the at least one field, and monitor the returned query results; when a query result is detected, output the query result returned by the current level query, and stop querying the target keywords of the next level field in the at least one field.

[0088] In another possible implementation, the aforementioned query unit 330 is specifically used to: perform a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of at least one field, and monitor the returned query results; when a query result is detected, repeat the step of querying the target keywords of the next level field in the at least one field based on the query result of the current level query, until no search results are returned, and output the query result returned by the previous level query.

[0089] In another possible implementation, the aforementioned determining unit 320 is further configured to: query the initial priority corresponding to at least one field, and determine whether there are multiple target fields among the at least one field, wherein the initial priorities of the multiple target fields are equal; if so, determine the relative priority among the multiple target fields according to user preferences; and determine the priority among at least one field according to the initial priority corresponding to at least one field and the relative priority among the multiple target fields.

[0090] In another feasible implementation, the aforementioned determining unit 320 is specifically used to: count the click component of each target field and the total click volume of peer fields, wherein peer fields are all fields in the corpus that have the same initial priority as multiple target fields; calculate the ratio of the click component of each target field to the total click volume to obtain the click percentage of each target field, and compare the numerical relationship between the click percentages of each target field; and determine the relative priority among multiple target fields according to the numerical relationship.

[0091] In another feasible approach, when there are target fields with the same click percentage among multiple target fields, the aforementioned determining unit 320 is further used to: determine the order of target fields with the same click percentage according to the position order of the target keywords under each target field in the complex statement; and determine the relative priority of target fields with the same click percentage according to the order.

[0092] In another possible implementation, before determining that a complex statement has been received, the determining unit 320 is further configured to: after receiving the user's query statement, perform a query on the query statement; if no query results are returned, detect the number of target keywords in the query statement; if the number of target keywords in the query statement is detected to be greater than or equal to two, determine that the query statement is a complex statement and determine that a complex statement has been received.

[0093] This application also provides a query device, see [link to relevant documentation] Figure 4 As shown in the figure, the query device in this embodiment may include a processor 410, a transceiver 430, and a memory 430. The processor 410, transceiver 430, and memory 430 are connected via a bus 440. The processor 410 is used to execute multiple instructions; the transceiver 430 is used to interact with other devices; and the memory 430 is used to store multiple instructions suitable for loading and execution by the processor 410, as described in the above embodiment, for complex querying in a vehicle-in-the-interior scenario.

[0094] The processor 410 can be an Electronic Control Unit (ECU), a Central Processing Unit (CPU), a general-purpose processor, a coprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The processor 410 can also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a 5SP and a microprocessor, etc. In this embodiment, the processor 410 can be a microcontroller. By programming the microcontroller, various control functions can be implemented. The processor has the advantages of powerful computing capabilities and fast processing speed. Specifically, the processor 410 executes the functions of the parsing unit 310, which, upon determining that a complex statement has been received, parses out multiple target keywords from the complex statement; it also executes the functions of the determining unit 320, which queries the fields to which each target keyword belongs to obtain at least one field; and it executes the functions of the query unit 330, which sets the query relationship between target keywords under the same field to OR, and performs a step-by-step query on the multiple target keywords according to the priority of the aforementioned at least one field, so as to output at least one query result. The transceiver 420 receives the user's query statement and outputs the query result.

[0095] In another possible implementation, the processor 410 is specifically configured to: perform a step-by-step query on the target keywords under each field of at least one field in descending order of priority of at least one field, and monitor the returned query results; when a query result is detected, output the query result returned by the current level query, and stop querying the target keywords of the next level field of at least one field.

[0096] In another possible implementation, the processor 410 is specifically used to: perform a step-by-step query on the target keywords under each field in the at least one field in descending order of priority of the at least one field, and monitor the returned query results; when a query result is detected, repeat the step of querying the target keywords of the next level field in the at least one field based on the query result of the current level query, until no search results are returned, and output the query result returned by the previous level query.

[0097] In another possible implementation, the processor 410 is further configured to: query the initial priority corresponding to at least one field, and determine whether there are multiple target fields among the at least one field, wherein the initial priorities of the multiple target fields are equal; if so, determine the relative priority among the multiple target fields according to user preference; and determine the priority among at least one field according to the initial priority corresponding to at least one field and the relative priority among the multiple target fields.

[0098] In another possible implementation, the processor 410 is specifically used to: count the click component of each target field and the total click volume of peer fields, wherein peer fields are all fields in the corpus that have the same initial priority as multiple target fields; calculate the ratio of the click component of each target field to the total click volume to obtain the click percentage of each target field, and compare the numerical relationship between the click percentages of each target field; and determine the relative priority among multiple target fields according to the numerical relationship.

[0099] In another feasible approach, when there are target fields with the same click percentage among multiple target fields, the processor 410 is further configured to: determine the order of target fields with the same click percentage according to the position order of the target keywords under each target field in the complex statement; and determine the relative priority of target fields with the same click percentage according to the order.

[0100] In another possible implementation, before determining that a complex statement has been received, the processor 410 is further configured to: after receiving the user's query statement, perform a query on the query statement; if no query results are returned, detect the number of target keywords in the query statement; if the number of target keywords in the query statement is detected to be greater than or equal to two, determine that the query statement is a complex statement and determine that a complex statement has been received.

[0101] In one embodiment, this application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor and executing the methods in any of the foregoing embodiments. A processor is used to execute the plurality of instructions; a memory is used to store the plurality of instructions, which are loaded by the processor and executed as in the above embodiments, representing a complex statement query method applied to an in-vehicle scenario.

[0102] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0103] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A complex sentence query method applied to an in-vehicle scene, characterized in that, The method comprises the following steps: After determining that a complex sentence is received, a plurality of target keywords in the complex sentence are parsed; Each target keyword is queried to belong to a field to obtain at least one field, wherein the field is set according to different query fields in a vehicle scene; The query relationship between the target keywords in the same field is set as OR, and the plurality of target keywords are sequentially queried according to the priority between the at least one field to output at least one query result; The priority between the at least one field is determined by the following steps: The initial priority of the field is determined according to the importance or function level of the field; If there are a plurality of target fields with equal initial priority in the at least one field, the relative priority between the plurality of target fields is determined according to the user click volume of the plurality of target fields and / or the position order of the target keywords under the target field in the complex sentence; The priority between the at least one field is determined according to the initial priority corresponding to the at least one field and the relative priority between the plurality of target fields.

2. The method of claim 1, wherein, The sequential query of the plurality of target keywords according to the priority between the at least one field to output at least one query result comprises: The target keywords under each field in the at least one field are sequentially queried according to the priority of the at least one field from high to low, and the returned query result is monitored; When the query result is monitored, the query result returned by the current query is output, and the query of the target keywords of the next level field in the at least one field is stopped.

3. The method of claim 1, wherein, The sequential query of the plurality of target keywords according to the priority between the at least one field to output at least one query result comprises: The target keywords under each field in the at least one field are sequentially queried according to the priority of the at least one field from high to low, and the returned query result is monitored; When the query result is monitored, the query of the target keywords of the next level field in the at least one field is repeated on the basis of the query result of the current query, and when no query result is returned, the query result returned by the previous query is output.

4. The method of claim 1, wherein, When the query result is output, the priority of the output mode is adjusted according to the current vehicle information or the user habit.

5. The method of claim 1, wherein, The relative priority between the plurality of target fields is determined according to the user click volume of the plurality of target fields, which comprises: The click volume of each target field and the total click volume of the same level field are counted, wherein the same level field is all fields in a corpus with the same initial priority as the plurality of target fields; The ratio of the click volume of each target field to the total click volume is calculated to obtain the click proportion of each target field, and the numerical size relationship between the click proportions of each target field is compared; The relative priority between the plurality of target fields is determined according to the numerical size relationship.

6. The method of claim 5, wherein, In the case that there are target fields with the same click proportion in the plurality of target fields, the method further comprises: determine the order among the target fields with consistent click-through rate according to the position order of the target keywords in the complex sentence under the target fields; determine the relative priority among the target fields with consistent click-through rate according to the order.

7. The method of claim 1, wherein, Before determining that the complex sentence is received, the method further comprises: after receiving the query sentence of the user, query the query sentence; if there is no query result returned, detect the number of target keywords of the query sentence; if it is detected that the number of target keywords of the query sentence is greater than or equal to two, it is determined that the query sentence is a complex sentence, and it is determined that the complex sentence is received.

8. A query device, characterized in that The query device comprises: a parsing unit configured to parse a plurality of target keywords in the complex sentence after determining that the complex sentence is received; a determination unit configured to query at least one field to which each target keyword belongs, wherein the field is set according to different query domains under the in-vehicle scene; a query unit configured to set the query relationship between the target keywords under the same field as OR, and perform step-by-step query on the plurality of target keywords according to the priority among the at least one field to output at least one query result; The priority among the at least one field is determined by the following steps: determine the initial priority of the field according to the importance or function level of the field; if there are a plurality of target fields with equal initial priority in the at least one field, determine the relative priority among the plurality of target fields according to the user click-through volume of the plurality of target fields and / or the position order of the target keywords in the complex sentence under the target fields; determine the priority among the at least one field according to the initial priority corresponding to the at least one field respectively, and the relative priority among the plurality of target fields.

9. A query device, characterized in that The query device comprises a processor, a transceiver and a memory, the processor and the memory are connected through a bus; the processor is used to execute a plurality of instructions; the transceiver is used to interact with other devices; the memory is used to store the plurality of instructions, the instructions are suitable for being loaded and executed by the processor to execute the complex sentence query method in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a plurality of instructions, the instructions are suitable for being loaded and executed by the processor to execute the complex sentence query method in any one of claims 1 to 7.

Images