Field staff visiting duration recording system and method

Abstract

The invention relates to a data statistical analysis technology, discloses an outworker visiting duration recording system and method, and solves the problems that a visiting duration recording scheme in the prior art is inaccurate in data, insufficient in integrity and incapable of effectively verifying the authenticity of visiting. In the scheme of the invention, field staff punch in and record the starting and ending time of the visit through the system, so that the theoretical visit duration is calculated, the system extracts the high-frequency keyword according to the task content of the visit task, and in the visit process, the system continuously records the audio information and records the occurrence frequency and position of the high-frequency keyword in the visit process; determining a reasonable effective duration interval of the visiting in combination with business experience and historical data; and determining the final visit duration according to the comparison between the theoretical visit duration and the effective duration interval, and verifying the authenticity of the visit based on the audio information recorded in the visit process. The method is suitable for checking attendance of field staff of enterprises.

Description

Technical Field

[0001] The present invention relates to data statistical analysis technology, and particularly to a system and method for recording the visit duration of field staff. Background Art

[0002] During the work process of field staff, due to their strong mobility and the characteristic of being unable to work at a fixed workstation for a long time, the traditional monitoring methods using fixed devices such as cameras are obviously not applicable. With the progress of technology, there are already various means in the prior art, such as mobile monitoring software, clock-in mechanisms, and working hour management software, etc., to monitor the actions and visit durations of field staff to ensure work efficiency and accuracy, which are specifically introduced as follows:

[0003] (1) Mobile monitoring software

[0004] That is, by installing specific monitoring software on the mobile devices of field staff, their locations, movement trajectories, and stay times can be tracked in real time. Such software usually has the following functions:

[0005] Location tracking: Obtain the geographical location information of field staff in real time, whether through GPS, base station positioning, or Wi-Fi positioning, so as to record the location information of field staff.

[0006] Trajectory recording: Automatically generate the movement trajectory map of field staff according to the positioning information. This not only helps enterprises understand the actual movement routes of field staff, but also can be used as evidence for work or dispute resolution when necessary.

[0007] Stay duration statistics: Record the stay times of field staff at various locations, so as to calculate their action visit durations.

[0008] (2) Clock-in mechanism

[0009] The clock-in mechanism is another common method for monitoring the working hours of field staff. Although traditional clock-in machines are not applicable to field staff, modern technology has provided more flexible clock-in methods, such as remote clock-in through mobile applications, text messages, or phone calls, etc. Field staff can record clock-in information through mobile applications after arriving at the visit location or completing the visit task. This method is not only convenient and fast, but also can upload clock-in information to the enterprise background in real time. For field staff who cannot use mobile applications, the enterprise can provide text message or phone call clock-in services. Field staff only need to send text messages or make phone calls according to the specified format to complete clock-in.

[0010] (3) Working hour management software

[0011] The working hours management software can also be used to monitor the working hours of field staff. Such software allows employees to record their working hours online and automatically generates reports. For field staff, they can record their visit duration anytime and anywhere through mobile devices such as mobile phones.

[0012] Although the above methods solve the problem of monitoring the working hours of field staff to a certain extent, there are still some deficiencies. For example, the mobile monitoring software relies heavily on the positioning function. In an environment with poor signal or many obstacles, there will be positioning deviation, which affects the accurate calculation of the visit duration; the clock-in mechanism depends on the conscious compliance of field staff. If field staff forget to clock in, deliberately do not clock in, or clock in early / delayed, it will lead to inaccurate data; the working hours management software needs to manually input to record the working hours, and there are also inaccurate data caused by human factors. In addition, several existing solutions only record the start and end time nodes of the visit, do not match the task content with the visit duration, the data integrity is insufficient, and the authenticity of the visit cannot be effectively verified. Summary of the Invention

[0013] The technical problem to be solved by the present invention is: to propose a system and method for recording the visit duration of field staff, and solve the problems of inaccurate data, insufficient integrity, and inability to effectively verify the authenticity of the visit in the existing visit duration recording solutions.

[0014] The technical solution adopted by the present invention to solve the above technical problems is:

[0015] On the one hand, the present invention provides a system for recording the visit duration of field staff, including:

[0016] A clock-in record module for recording the start time and end time of the visit of field staff;

[0017] A task assignment module for assigning visit tasks or allowing field staff to plan and enter visit tasks by themselves, and extracting high-frequency keywords in the task content;

[0018] An audio recording module for recording audio information during the visit;

[0019] An audio processing module for extracting feature information from the recorded audio information, and recording the frequency and position of the high-frequency keywords extracted from the task content during the visit;

[0020] A duration judgment module for calculating the theoretical visit duration according to the recorded start time and end time of the visit; and determining a reasonable effective duration interval for this visit according to the frequency and position of the high-frequency keywords during the visit, combined with business experience and historical data; and,

[0021] Compare the theoretical visit duration with the reasonable effective duration range of this visit. If the theoretical visit duration is within the effective duration range, then use the theoretical visit duration as the final visit duration; otherwise, determine the final visit duration within the effective duration range through interpolation method;

[0022] A visit verification module, which is used to verify the authenticity of this visit based on the audio information recorded during the visit. If the verification passes, mark this visit as true and valid; otherwise, mark the authenticity of this visit as doubtful.

[0023] Furthermore, the field staff visit duration recording system is configured as a mobile office application integrating GPS positioning, online / offline clock-in, task assignment, progress reporting, and customer communication functions.

[0024] Furthermore, the visit verification module is specifically used for: extracting audio features according to the recorded audio information, comparing the extracted audio features with the pre-stored characteristic audio to determine the audio category, and the audio category includes on-site real human voices and other sounds; the extracted audio features include: sound source distance, decibel value, and sound source direction;

[0025] According to the determined audio category, construct a sound map based on the audio features belonging to the on-site real human voice category;

[0026] According to the sound map, screen out the audio information of the sound source within a certain range from the field staff;

[0027] Judge whether the screened audio information belongs to the visited person. If so, continue with the audio authenticity verification; otherwise, determine that the authenticity of this visit is doubtful.

[0028] Furthermore, the audio authenticity verification includes:

[0029] Compare the time stamp of the audio file with the time of the clock-in record to verify whether the audio was recorded within the visit time period; at the same time, analyze the background noise in the audio and compare it with the typical background noise of the visit location to verify the authenticity of the audio.

[0030] Or, the visit verification module is specifically used for: extracting audio features according to the recorded audio information, comparing the extracted audio features with the pre-stored characteristic audio to determine the audio category, and the audio category includes on-site real human voices and other sounds; the extracted audio features include: sound source distance, decibel value, and sound source direction;

[0031] Construct a sound source-distance function image with time as the abscissa and sound source distance as the ordinate;

[0032] The audio information of each sound source extracted is plotted on the image to form multiple sound source distance curves, and each sound source distance curve represents the change of the sound source distance of a sound source over time;

[0033] By analyzing the sound source-distance function image, the sound source distance curves of the field staff and the visited person are matched with the reasonable effective duration interval of this visit. If the match is successful and there are no abnormal fluctuations in the sound source distance curve within the reasonable effective duration interval of this visit, it is determined that this visit is truly effective; otherwise, it is determined that the authenticity of this visit is in doubt.

[0034] On the other hand, based on the above system, the present invention also provides a method for recording the visit duration of field staff, including the following steps:

[0035] S1. The field staff obtains a visit task, and the system extracts high-frequency keywords according to the task content of the visit task;

[0036] S2. After the field staff arrives at the visit destination, they punch in through the system, and the system records the visit start time;

[0037] S3. During the visit, the system continuously records audio information and records the frequency and position of the high-frequency keywords extracted from the task content during the visit;

[0038] S4. After the visit ends, the field staff punches in through the system again to record the visit end time;

[0039] S5. The system calculates the theoretical visit duration according to the recorded visit start time and visit end time;

[0040] S6. The system determines the reasonable effective duration interval of this visit according to the frequency and position of the high-frequency keywords during the visit, combined with business experience and historical data;

[0041] S7. The system compares the theoretical visit duration with the reasonable effective duration interval of this visit. If the theoretical visit duration is within the effective duration interval, the theoretical visit duration is used as the final visit duration; otherwise, the final visit duration is determined within the effective duration interval by the interpolation method;

[0042] S8. The system verifies the authenticity of this visit based on the audio information recorded during the visit. If the verification passes, this visit is marked as truly effective; otherwise, this visit is marked as having doubtful authenticity.

[0043] Further, in step S1, the method for the system to extract high-frequency keywords according to the task content of the visit task includes:

[0044] After the system extracts keywords from the task content using NLP technology, it performs word frequency statistics and sorting on the keywords, and filters out high-frequency keywords according to business requirements.

[0045] Further, in step S8, the system verifies the authenticity of this visit based on the audio information recorded during the visit. One of the methods used includes:

[0046] According to the recorded audio information, audio features are extracted, and the extracted audio features are compared with the pre-stored characteristic audio to determine the audio category. The audio categories include on-site real human voices and other sounds; the extracted audio features include: sound source distance, decibel value, and sound source direction.

[0047] According to the determined audio category, a sound map is constructed based on the audio features belonging to the on-site real human voice category.

[0048] According to the sound map, the audio information of the sound sources within a certain range from the field staff is filtered out.

[0049] Judge whether the filtered audio information belongs to the person being visited. If so, continue with the audio authenticity verification; otherwise, determine that the authenticity of this visit is in doubt.

[0050] Further, the audio authenticity verification includes:

[0051] Compare the timestamp of the audio file with the time of the clock-in record to verify whether the audio was recorded within the visit time period; at the same time, analyze the background noise in the audio and compare it with the typical background noise of the visit location to verify the authenticity of the audio.

[0052] Further, in step S8, the system verifies the authenticity of this visit based on the audio information recorded during the visit. Another method used includes:

[0053] According to the recorded audio information, audio features are extracted, and the extracted audio features are compared with the pre-stored characteristic audio to determine the audio category. The audio categories include on-site real human voices and other sounds; the extracted audio features include: sound source distance, decibel value, and sound source direction.

[0054] Taking time as the abscissa and sound source distance as the ordinate, construct a sound source-distance function image.

[0055] Plot the audio information of each extracted sound source on this image to form multiple sound source distance curves. Each sound source distance curve represents the change of the sound source distance of a sound source over time.

[0056] By analyzing the sound source - distance function image, the sound source distance curve between the field staff and the visited person is matched with the reasonable effective duration interval of this visit. If the match is successful and there are no abnormal fluctuations in the sound source distance curve within the reasonable effective duration interval of this visit, then this visit is determined to be truly effective; otherwise, the authenticity of this visit is suspected.

[0057] The beneficial effects of the present invention are as follows:

[0058] (1) By deeply analyzing the task content and extracting high - frequency keywords, it is beneficial to quickly understand the core theme and key points of the task. According to the appearance position and frequency of the keywords, combined with business experience and historical data, a reasonable effective duration interval is judged. When determining the visit duration, the theoretical visit duration recorded by the clock - in record is compared with this reasonable effective duration interval to comprehensively determine the final visit duration, thereby improving the accuracy and rationality of the visit duration.

[0059] (2) Based on the audio information record during the visit and matching it with the keywords re - extracted from the visit task, the visit duration is thus matched with the visit task, enriching the data record content of the visit and making the data integrity of the entire visit higher.

[0060] (3) By means of constructing a sound map, sound source - distance function image, etc. to verify the authenticity of the audio information recorded during the visit, and marking and outputting the situations where the authenticity of the visit is suspected, it is beneficial to quickly discover potential cheating behaviors. Brief Description of the Drawings

[0061] Figure 1 It is the block diagram of the system for recording the visit duration of field staff in the present invention;

[0062] Figure 2 It is the schematic flow chart of the method for recording the visit duration of field staff in the present invention. Detailed Embodiments

[0063] The present invention aims to provide a system and method for recording the visit duration of field personnel, which solves the problems of inaccurate data, insufficient integrity, and inability to effectively verify the authenticity of visits in the existing visit duration recording solutions. Its core idea is: in the process of determining the visit duration, the concept of "reasonable effective duration interval" is introduced. This duration interval is a reasonable interval calculated by combining business experience and historical data based on the high-frequency keywords extracted through in-depth analysis of the visit tasks, the frequency and position of their occurrences during the visit. By comparing the theoretical visit duration with the reasonable interval, the final visit duration is comprehensively determined, which can improve the accuracy and reasonableness of visits and avoid the problem of false reporting of visit durations in traditional technologies; by recording the audio information (excluding the audio content) during the visit and matching it with the keywords extracted from the visit tasks, the visit duration is matched with the visit tasks, enriching the data record content of the visit and increasing data integrity; in addition, the present invention also verifies the authenticity of the audio information recorded during the visit by constructing a sound map, a sound source-distance function image, etc., marking the situations where the authenticity of the visit is in doubt, which is conducive to quickly discovering potential cheating behaviors.

[0064] Specifically, the structure of a system for recording the visit duration of field personnel provided by the present invention is shown in Figure 1 , and it includes the following functional modules:

[0065] The check-in record module is used to record the start time and end time of the visit of field personnel;

[0066] The task assignment module is used to assign visit tasks or allow field personnel to plan and enter visit tasks by themselves, and extract high-frequency keywords from the task content;

[0067] The audio recording module is used to record audio information during the visit;

[0068] The audio processing module is used to extract feature information from the recorded audio information, and record the frequency and position of the high-frequency keywords extracted from the task content during the visit;

[0069] The duration judgment module is used to calculate the theoretical visit duration according to the recorded start time and end time of the visit; and determine the reasonable effective duration interval of this visit according to the frequency and position of the high-frequency keywords during the visit, combined with business experience and historical data; and,

[0070] Compare the theoretical visit duration with the reasonable effective duration interval of this visit. If the theoretical visit duration is within the effective duration interval, then use the theoretical visit duration as the final visit duration; otherwise, determine the final visit duration within the effective duration interval by interpolation method;

[0071] A visit verification module is used to verify the authenticity of the current visit based on the audio information recorded during the visit. If the verification passes, the current visit is marked as true and valid; otherwise, the authenticity of the current visit is marked as doubtful.

[0072] In specific implementation, the above system can be configured as a mobile office application integrating functions such as GPS positioning, online / offline clock-in, task assignment, progress reporting, and customer communication. Ensure that the system has a friendly interface and is easy to operate, facilitating field staff to quickly get started.

[0073] The following specifically describes the method for recording the visit duration of field staff implemented based on the above system. Refer to Figure 2 , and this method includes the following steps:

[0074] S1. Obtain a visit task and extract high-frequency keywords according to the task content:

[0075] In this step, the manager assigns visit tasks to field staff through the system or allows field staff to plan tasks in the system by themselves. Before departure, field staff briefly input or confirm the specific content of the visit through the system, including customer name, address, appointment time, visit purpose, expected goals, etc.

[0076] The system automatically uses NLP technologies such as word segmentation, part-of-speech tagging, and named entity recognition to deeply analyze the task content according to the task content input by field staff, and extract potential keywords; subsequently, through clustering analysis, similar keywords are grouped into one category to better understand the theme and key points of the task content; by establishing a keyword dictionary containing common product names, customer industry terms, visit purpose phrases, etc., based on dictionary matching, keywords related to the task content can be quickly extracted; then, word frequency statistics are performed on the extracted keywords to find the keywords with higher frequencies of occurrence. These high-frequency keywords usually reflect the core theme and key points of the task content; finally, the keywords are sorted according to the word frequency statistics results. At the same time, according to business requirements and experience, the keywords most relevant and valuable to the visit task are selected as the high-frequency keywords determined in this step.

[0077] S2. Field staff clock in, and the system records the start time of the visit:

[0078] In this step, after arriving at the customer's location, field staff clock in through the mobile application. The system records the start time and re-confirms the visit task content. When clocking in, the system needs to automatically record the longitude and latitude information of the clock-in location to verify the actual location of field staff. The mobile application provides online or offline clock-in functions, thus ensuring that attendance can be recorded even in places with poor signals.

[0079] S3. The system records the audio information during the visit and determines the frequency and location of the high-frequency keywords during the visit:

[0080] In this step, during the visit, the field staff conducts product introductions, customer needs discussions, etc. according to the plan. At the same time, the system continuously records audio information through the microphone of the field staff's mobile device. These audio information contain all the voice communications within the visit area where the field staff is located, but do not contain specific content information to protect privacy. During the recording process, the system needs to preprocess the audio information, such as noise reduction, echo cancellation, etc., to improve the audio quality.

[0081] During this period, the system analyzes the location (such as the opening remarks, discussion content, closing remarks, etc.) and frequency of the high-frequency keywords extracted in step S1 based on the recorded audio information, for use in determining a reasonable effective duration range for the visit subsequently.

[0082] S4. Punch the card again after the visit ends to record the end time of the visit:

[0083] In this step, after the visit ends, before the field staff leaves the customer location, punch the card again through the mobile application, and the system records the end time of the visit.

[0084] S5. Calculate the theoretical visit duration;

[0085] In this step, the system can calculate a visit duration based on the recorded start time and end time of the visit. This duration is the time interval between the two punches for the start and end of the visit, which is called the "theoretical visit duration" in the present invention.

[0086] S6. Calculate the reasonable effective duration range for this visit:

[0087] In this step, the system determines the reasonable effective duration range for this visit based on the frequency and location of the high-frequency keywords during the visit, combined with business experience and historical data.

[0088] In an exemplary solution, the occurrence of each keyword can be regarded as a "spot", and its position, size, and color can be adjusted according to the importance, occurrence frequency, and duration of the keyword. The denser and darker (or larger) the spots are, the higher the occurrence frequency of the keywords in that time period, indicating the occurrence of effective communication. Therefore, the effective communication range can be determined according to the distribution of spots in the sound map. The effective communication range (i.e., the effective duration range) can be defined as the time period when the spot density exceeds a certain threshold. Among them, the threshold can be adjusted according to historical data, business experience, and actual tests. During the whole process, only when the keyword appears, the system automatically records the frequency and position of the keyword, and does not record the specific communication content.

[0089] S7. Determine the final visit duration by comparing the theoretical visit duration with the effective duration range:

[0090] In this step, the system compares the theoretical visit duration with the reasonable effective duration range of this visit. If the theoretical visit duration is within the effective duration range, the theoretical visit duration is used as the final visit duration; otherwise, the final visit duration is determined within the effective duration range by interpolation. Among them, the interpolation method can be selected according to specific circumstances, such as linear interpolation, Lagrange interpolation, etc.

[0091] S8. Verify the authenticity of this visit based on the audio information recorded during the visit:

[0092] In this step, the system verifies the authenticity of this visit based on the audio information recorded during the visit. If the verification passes, this visit is marked as true and valid; otherwise, the authenticity of this visit is marked as doubtful.

[0093] Specifically, there are two ways to verify authenticity. One is to verify through the sound map, and the other is to verify by constructing the sound source - distance function image. The following is a specific description of each:

[0094] 1. Verification through the sound map:

[0095] Extract the audio features from the recorded audio information, including: sound source distance, decibel value, sound source direction, etc. These feature information can reflect the relative position and sound intensity between the field staff and the visited person, providing key data for constructing the sound map; and compare the extracted audio features with the pre - stored feature audio to determine the audio category, where the audio category includes on - site real human voices and other sounds (such as environmental sounds or human voices in players, etc.).

[0096] Then, use the extracted feature information, especially the sound source distance, to layer the audio information according to the distance between the sound source and the field staff. Each layer represents a distance range and is represented by different colors or marks. The sound map can intuitively display the relative position changes between the field staff and the visited person during the visit. This helps to determine whether the field staff is indeed having a face - to - face conversation with the customer.

[0097] In the sound map, filter out the audio information where the distance between the sound source and the field staff changes within a certain range. This range can be set according to the actual situation to reflect the normal face-to-face communication distance range, that is, the audio information is emitted by the visited person. During a general normal visit, the distance between the field staff and the visited person will remain within a range, that is, it fluctuates within a range. By filtering, audio information where the sound source is too far or too close to the field staff can be excluded, and this information indicates that the field staff is not having a face-to-face communication with the visited person.

[0098] For the filtered audio information, further analyze the relative position of the sound source distance and the field staff to determine whether this audio information actually comes from the visited person. If so, continue with the audio authenticity verification. Otherwise, the authenticity of this visit is suspected.

[0099] Continuing with the audio authenticity verification can further verify the authenticity of the visit from aspects such as timestamp comparison and scene matching. The main purpose is to determine whether the audio is a real-time audio recorded during the visit period, rather than pre-recorded by the field staff. In an exemplary solution, the verification method is as follows:

[0100] Timestamp comparison: The timestamp of the audio file can be compared with the time of the punch record to verify whether the audio was recorded during the visit period.

[0101] Scene matching: That is, by analyzing the background noise in the audio, such as traffic noise, environmental noise, etc., and comparing it with the typical background noise of the visit location to verify the authenticity of the audio. The visit location is extracted from the visit task content obtained by the field staff. Through the visit location, the scene where the field staff is located during the visit can be initially judged (such as the floor area, surrounding environment, shape, structure, decoration style, etc., and the sound characteristics generated). Compare the scene characteristics in the audio with the actual characteristics of the visit location and analyze the degree of matching between them. If the scene characteristics in the audio are highly consistent with the actual characteristics of the visit location, it further verifies the authenticity and scene matching degree of the audio.

[0102] 2. Sound source - distance function image verification:

[0103] First, based on the audio characteristics in the audio information, construct a sound source - distance function image, that is, use time as the abscissa and the sound source distance as the ordinate to construct a sound source - distance function image. Plot the audio information of each extracted sound source (including customers, environmental sounds, other staff in the customer's company, etc.) on this image to form multiple curves, and each curve represents the change of the sound source distance of a sound source over time.

[0104] Next, by analyzing the sound source - distance function image, the sound source distance curve between the field staff and the visited person can be obtained, which can reflect the relative position change between the field staff and the customer. During a normal visit, the sound source distance curve between the field staff and the customer will fluctuate within a relatively stable range, reflecting their face - to - face communication situation. If the sound source distance curve shows abnormal fluctuations (such as suddenly increasing or decreasing to an unreasonable range), it indicates that the field staff is not really having a face - to - face communication with the customer.

[0105] Finally, combining the analysis results of the sound source - distance function image and the effective communication periods previously judged and obtained by the system, comprehensively judge the authenticity and effectiveness of the visit: If the sound source - distance function image matches the effective communication period and there are no abnormal fluctuations, the visit can be considered real and effective. If the sound source - distance function image does not match the effective communication period or there are abnormal fluctuations, further investigation is required or it is marked as an invalid visit.

[0106] Embodiment

[0107] This embodiment takes the visit of field staff Li Ming to retail chain store A located in the city center as an example.

[0108] Before Li Ming conducts the visit, he inputs the visit task content through the mobile office application, including the customer name (large retail chain store), address, appointment time, visit purpose (new product promotion and sales strategy discussion), and the expected goal to be achieved (signing a sales contract or reaching a preliminary cooperation intention). The system extracts high - frequency keywords such as "new product", "sales strategy", "retail chain store", etc. according to the task content input by Li Ming using NLP technology.

[0109] After Li Ming arrives at the customer's location, he uses the mobile office application to punch in and record the start time of the visit.

[0110] During the visit, when Li Ming communicates with the customer, he uses the recording function in the mobile office application to record the high - frequency keywords that appear during the discussion in real - time (instead of recording all the audio). These audio information (high - frequency keywords) are automatically uploaded to the system for subsequent analysis.

[0111] After the visit, Li Ming clocked in for the end of the visit again through the mobile office application. At the same time, the system judged a reasonable effective duration interval based on the appearance position and frequency of the keywords, and compared it with the actual visit duration. For example, Li Ming started the visit at 14:00 and ended at 15:00. The total visit duration clocked in by Li Ming was 1 hour. During this 1 hour, high-frequency keywords such as "new product" and "sales strategy" appeared in the time period from 14:05 to 14:50. The system combined business experience, historical data, and the appearance position of these high-frequency keywords to judge a reasonable effective duration interval, such as 50 minutes - 60 minutes. The system compared the visit duration of 60 minutes recorded by Li Ming with the effective duration interval (50 minutes - 60 minutes) and found that it was within the interval, so it directly used this duration (60 minutes) as the final visit duration.

[0112] To verify the authenticity and effectiveness of the visit, the system screened and verified the audio information within the effective duration interval. The system obtained all the audio information from the start of Li Ming's clock-in to the end of the clock-in. By comparing with the pre-stored characteristic audio, the system determined the category of the audio.

[0113] Before classifying the audio, the system needs to establish a database containing characteristic audio of various categories. These characteristic audios are carefully selected and labeled, and are audio samples that can represent the typical characteristics of various categories. Each sample in the characteristic audio library contains a series of extracted characteristics, which are the spectral characteristics, rhythm characteristics, timbre characteristics, etc. of the audio.

[0114] For the audio signal to be classified, the system will first perform preprocessing, including operations such as denoising, filtering, and framing, to extract a purer and more representative audio signal. Next, the system will extract the characteristics of the audio signal to be classified, extracting the same type of characteristics as those in the pre-stored characteristic audio library. These characteristics can be Mel Frequency Cepstral Coefficients (MFCC), Linear Predictive Cepstral Coefficients (LPCC), spectral centroid, etc. After extracting the characteristics, the system will match the characteristics of the audio to be classified with the characteristics in the pre-stored characteristic audio library. This matching process is achieved by calculating the similarity between the characteristics (such as Euclidean distance, cosine similarity, etc.). According to the calculation result of the similarity, the system will assign the audio to be classified to the category to which the most similar characteristic audio belongs. In this way, the audio information is divided into two categories: human voice and other sounds (ambient sounds or human voices in the player).

[0115] Then, the characteristic information in the audio information belonging to the human voice is extracted to construct a sound map, reflecting the distance between each audio information and Li Ming during this period.

[0116] The system filters out audio information with distance changes within a certain range and determines whether this audio belongs to the visited person. For example, the audio information V with a distance change within 2 meters from Li Ming during this period is filtered out. If the audio information V reflects a distance of 1 meter to 2 meters from Li Ming on the sound map, it is determined that the audio information V belongs to the visited person; otherwise, it does not belong to the visited person, and this period of time is marked as invalid.

[0117] If the audio information V belongs to the visited person, it is necessary to conduct a second audio verification on it. The system also further verifies the authenticity of the audio through timestamp comparison and background noise analysis. For example, the timestamp of the audio file is compared with the timestamp of the punch card record to verify whether the audio was recorded during the visiting period; at the same time, the background noise in the audio is analyzed and compared with the typical background noise of the visiting location to verify the authenticity of the audio.

[0118] Finally, it should be noted that the above embodiments are only preferred embodiments and do not limit the present invention. It should be pointed out that for those of ordinary skill in the art of this technology, without departing from the spirit and scope of the present invention as protected by the claims, several modifications, equivalent replacements, improvements, etc. can be made, and all of them should be included in the protection scope of the present invention.

Claims

1. A field staff visit duration recording system, characterized in that: include: The clock-in record module is used to record the start and end time of field staff's visits; The task assignment module is used to assign visiting tasks or allow field staff to plan and enter visiting tasks by themselves, and extract high-frequency keywords from the task content; Audio recording module, used to record audio information during the visit; An audio processing module, used to extract feature information from the recorded audio information, and record the frequency and position of the high-frequency keywords extracted from the task content during the visit; Duration determination module, used to calculate theoretical visit duration based on recorded visit start time and visit end time; And according to the frequency and location of high-frequency keywords in the visit process, combined with business experience and historical data, determine the reasonable effective duration range of this visit; as well as, Compare the theoretical visit duration with the reasonable effective duration interval of this visit. If the theoretical visit duration is within the effective duration interval, the theoretical visit duration is used as the final visit duration. Otherwise, the final visit duration is determined within the effective duration interval by interpolation method. The visit verification module is used to verify the authenticity of the visit based on the audio information recorded during the visit. If the verification is successful, the visit is marked as authentic and valid; otherwise, the authenticity of the visit is marked as questionable.

2. A field staff visit duration recording system as claimed in claim 1, characterized in that: The field staff visit duration recording system is configured as a mobile office application that integrates GPS positioning, online / offline clocking in, task allocation, progress reporting, and customer communication functions.

3. A field staff visit duration recording system as claimed in claim 1, characterized in that: The visit verification module is specifically used to: extract audio features according to the recorded audio information, and compare the extracted audio features with the pre-stored feature audio to determine the audio category, where the audio category includes real human voices and other sounds on site; The extracted audio features include: sound source distance, decibel value and sound source direction; According to the determined audio category, a sound map is constructed based on audio features belonging to the category of real human voice at the scene; According to the sound map, filter out the audio information of the sound source within a certain range from the field staff; Determine whether the filtered audio information belongs to the person being visited. If so, continue to verify the authenticity of the audio. Otherwise, determine that the authenticity of this visit is questionable.

4. A field staff visit duration recording system as claimed in claim 3, characterized in that: The audio authenticity verification includes: Compare the timestamp of the audio file with the time of the clock-in record to verify whether the audio was recorded during the visit time period; at the same time, analyze the background noise in the audio and compare it with the typical background noise of the visit location to verify the authenticity of the audio.

5. A field staff visit duration recording system as claimed in claim 1, characterized in that: The visit verification module is specifically used to: extract audio features according to the recorded audio information, and compare the extracted audio features with the pre-stored feature audio to determine the audio category, where the audio category includes real human voices and other sounds on site; The extracted audio features include: sound source distance, decibel value and sound source direction; With time as the horizontal axis and the distance from the sound source as the vertical axis, a sound source-distance function graph is constructed; The extracted audio information of each sound source is plotted on the image to form a plurality of sound source distance curves, each of which represents the change of the sound source distance of a sound source over time; By analyzing the sound source-distance function graph, the sound source distance curve between the field staff and the person being visited is matched with the reasonable effective time interval of this visit. If the match is successful and there is no abnormal fluctuation in the sound source distance curve within the reasonable effective time interval of this visit, then the visit is judged to be authentic and valid; otherwise, the authenticity of the visit is judged to be questionable.

6. A method for recording the duration of a field staff visit, applied to the field staff visit duration recording system as claimed in any one of claims 1 to 5, characterized in that: The method comprises the following steps: S1. Field staff obtains a visiting task, and the system extracts high-frequency keywords based on the content of the visiting task; S2. After arriving at the destination, the field staff clocks in through the system, and the system records the start time of the visit; S3. During the visit, the system continuously records audio information and records the frequency and location of the high-frequency keywords extracted from the task content during the visit; S4. After the visit, the field staff will clock in again through the system to record the end time of the visit; S5. The system calculates the theoretical visit duration based on the recorded visit start time and visit end time; S6. The system determines the reasonable effective duration of this visit based on the frequency and location of high-frequency keywords during the visit, combined with business experience and historical data; S7. The system compares the theoretical visit duration with the reasonable effective duration interval of this visit. If the theoretical visit duration is within the effective duration interval, the theoretical visit duration is used as the final visit duration. Otherwise, the final visit duration is determined within the effective duration interval by interpolation method. S8. The system verifies the authenticity of the visit based on the audio information recorded during the visit. If the verification is successful, the visit is marked as authentic and valid. Otherwise, the authenticity of the visit is marked as questionable.

7. A method for recording the duration of a field staff visit as claimed in claim 6, characterized in that: In step S1, the method for the system to extract high-frequency keywords according to the task content of the visit task includes: After the system extracts keywords from the task content using NLP technology, it counts and sorts the keywords, and selects high-frequency keywords based on business needs.

8. A method for recording the duration of a field staff visit as claimed in claim 6, characterized in that: In step S8, the system verifies the authenticity of the visit based on the audio information recorded during the visit, using the following methods: According to the recorded audio information, audio features are extracted, and the extracted audio features are compared with the pre-stored feature audio to determine the audio category, which includes real human voices and other sounds on the scene; the extracted audio features include: sound source distance, decibel value and sound source direction; According to the determined audio category, a sound map is constructed based on audio features belonging to the category of real human voice at the scene; According to the sound map, filter out the audio information of the sound source within a certain range from the field staff; Determine whether the filtered audio information belongs to the person being visited. If so, continue to verify the authenticity of the audio. Otherwise, determine that the authenticity of this visit is questionable.

9. A method for recording the duration of a field staff visit as claimed in claim 8, characterized in that: The audio authenticity verification includes: Compare the timestamp of the audio file with the time of the clock-in record to verify whether the audio was recorded during the visit time period; at the same time, analyze the background noise in the audio and compare it with the typical background noise of the visit location to verify the authenticity of the audio.

10. A method for recording the duration of a field staff visit as claimed in claim 6, characterized in that: In step S8, the system verifies the authenticity of the visit based on the audio information recorded during the visit, using the following methods: According to the recorded audio information, audio features are extracted, and the extracted audio features are compared with the pre-stored feature audio to determine the audio category, which includes real human voices and other sounds on the scene; the extracted audio features include: sound source distance, decibel value and sound source direction; With time as the horizontal axis and the distance from the sound source as the vertical axis, a sound source-distance function graph is constructed; The extracted audio information of each sound source is plotted on the image to form a plurality of sound source distance curves, each of which represents the change of the sound source distance of a sound source over time; By analyzing the sound source-distance function graph, the sound source distance curve between the field staff and the person being visited is matched with the reasonable effective time interval of this visit. If the match is successful and there is no abnormal fluctuation in the sound source distance curve within the reasonable effective time interval of this visit, then the visit is judged to be authentic and valid; otherwise, the authenticity of the visit is judged to be questionable.

Images