Express delivery front-end intelligent communication strategy optimization method, device, equipment and medium

Abstract

The application provides an express delivery pre-delivery intelligent communication strategy optimization method and device, equipment and medium, comprising: collecting user historical logistics interaction full data, processing through big data labeling, building a user personalized model containing answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics and complaint characteristics; real-time collection of package delivery dynamic data, time scene data and environment scene data to obtain real-time state data; based on the real-time state data, combining the user personalized model, dynamically adjusting the outbound mode and outbound timing; combining the adjusted outbound mode and outbound timing, and the AI voice language suitable for the user personalized characteristics and the current scene, generating an intelligent communication strategy for the current order, and initiating targeted delivery communication to the corresponding user.

Description

Technical Field

[0001] This invention relates to the field of logistics management technology, specifically to a method, apparatus, equipment, and medium for optimizing intelligent communication strategies before express delivery. Background Technology

[0002] With the digital and intelligent upgrade of the logistics industry, AI voice outbound calling has gradually replaced manual communication and become the mainstream way for users to communicate before express delivery. It is widely used in scenarios such as express arrival confirmation, delivery appointment, and collection point selection reminders, which greatly improves the communication efficiency of last-mile delivery.

[0003] Currently, most mainstream AI voice-based parcel delivery systems on the market employ fixed, template-based, and static operating rules, resulting in significant technical flaws and shortcomings in scenario adaptation: First, the call times are fixed, with voice calls initiated in batches according to a uniform preset time period. This lack of differentiation based on user habits, historical call behavior, and real-time scenario conditions makes it highly susceptible to calls made during non-calling periods such as work, rest, or travel, leading to a large number of rejected and unanswered invalid calls. Second, the communication methods are singular and rigid, with all users uniformly using a "voice call priority" mode, failing to differentiate between user SMS preferences, rejection characteristics, and other personalized needs, potentially causing user harassment. Third, the script templates are uniform and universal, lacking personalized adaptation. All users receive the same broadcast script, unable to link to users' historical parcel delivery selections or past communication feedback, resulting in a clunky interactive experience and low user cooperation and delivery confirmation conversion rates. Fourth, there is a lack of real-time scenario-based decision-making capabilities; communication strategies are not adjusted based on external real-time factors such as parcel delivery progress, weather, holidays, and road conditions, resulting in extremely poor scenario adaptability.

[0004] Existing technologies rely solely on simple, fixed program logic to execute outbound calls, lacking in-depth mining of user personalization characteristics, real-time scenario data fusion and analysis, and the ability to dynamically adapt and adjust strategies. This has long resulted in a series of problems, including a large volume of invalid outbound calls, wasted communication resources, low user connection rates, high complaint rates, and inefficient delivery and communication. The industry urgently needs an intelligent communication system that can achieve user personalization adaptation, real-time scenario linkage, and dynamic strategy optimization to address the many shortcomings of existing fixed-rule outbound calling technologies. Summary of the Invention

[0005] The main objective of this invention is to solve the technical problems in the prior art that rely solely on simple fixed program logic to perform outbound call operations, lacking in-depth mining of user personalized characteristics, fusion analysis of real-time scenario data, and dynamic strategy adaptive adjustment capabilities, resulting in a long-standing problem of large volume of invalid outbound calls, waste of communication resources, low user connection rate, high complaint rate, and low delivery and communication efficiency.

[0006] The first aspect of this invention provides a method for optimizing intelligent communication strategies before express delivery, comprising: Collect all historical logistics interaction data of users, and through big data tagging processing, construct a personalized user model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics; Real-time data collection of dynamic parcel delivery data, time-based data, and environmental data is used to obtain real-time status data. Based on real-time status data and combined with user-personalized models, the outbound calling method and timing are dynamically adjusted. By combining the adjusted outbound calling methods and timings with AI voice scripts tailored to user personalities and the current scenario, an intelligent communication strategy is generated for the current order, and targeted delivery communication is initiated to the corresponding user.

[0007] Optionally, in a first implementation of the first aspect of the present invention, the step of collecting all historical logistics interaction data of users and constructing a personalized user model including call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics through big data tagging processing includes: The data collected includes: user's historical package receipt records, AI voice call answering / rejection records, missed call feedback, human customer service complaint and suggestion records, historical collection point selection records, communication method preference records, and time period call pattern data. The collected full volume of user historical logistics interaction data is processed by cleaning, deduplication and normalization to obtain the processed full volume of user historical logistics interaction data. Based on the pre-processed full volume of user historical logistics interaction data, big data tagging is performed to generate any one or more user tags, including call answering feature tags, time period feature tags, communication preference feature tags, delivery method preference feature tags, and complaint feature tags. The call answering feature tags include: historical call answering frequency; The time-period feature tags include: rejection rates for morning, noon, evening, and night. The communication preference feature tags include: voice preference, SMS preference, and mixed communication preference; The delivery method preference feature tags include: pick-up preference and door-to-door delivery preference; The complaint feature tags include: outbound call harassment complaints and complaints of dissatisfaction with the script; A personalized user model is constructed based on the included call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics.

[0008] Optionally, in a second implementation of the first aspect of the present invention, when the delivery method preference feature label in the user's personalized model is collection preference, the collection point information with the highest probability of use associated with the current user is retrieved.

[0009] Optionally, in a third implementation of the first aspect of the present invention, the real-time collection of dynamic data on package delivery, time-based scene data, and environmental scene data includes: The dynamic data of package delivery includes: package category identification, remaining delivery time, overdue warning level, delivery delay duration, real-time distance between the delivery person and the delivery address, package status, and package abnormality mark. The time-based scenario data includes: weekday morning, noon, evening, and nighttime periods, and non-weekday morning, noon, evening, and nighttime periods; The environmental scene data includes: weather information and road condition information.

[0010] Optionally, in the fourth implementation of the first aspect of the present invention, the step of dynamically adjusting the outbound calling method and timing based on real-time status data and combined with a user personalization model comprises: dynamically adjusting the outbound calling method based on real-time collected dynamic data of package delivery and environmental scene data, combined with communication preference feature tags in the user personalization model, including: Based on real-time collected dynamic data on package delivery and environmental scene data, the current scenario is identified as either a Level 1 emergency scenario or a Level 2 routine scenario. When the current scenario is a Level 1 emergency scenario, and the communication preference feature label in the user's personalized model is the SMS preference feature label or the mixed communication preference feature label, then the call will be dynamically adjusted to prioritize voice calls. When the current scenario is a level 2 routine scenario, the corresponding communication preference feature labels in the user's personalized model are maintained; The first-level emergency scenarios include any one or more of the following: packages identified as priority delivery categories and non-receiveable categories, packages with time-limited signing, remaining delivery time less than a threshold, overtime warning level meeting preset requirements, delivery delays, real-time distance between the delivery person and the delivery address less than a threshold, packages in a pending delivery status, packages with abnormalities, extreme rain or snow meeting preset requirements in weather information, or road closures meeting preset requirements in road condition information. The secondary routine scenarios include scenarios other than the primary emergency scenarios.

[0011] Optionally, in the fifth implementation of the first aspect of the present invention, the step of dynamically adjusting the outbound calling method and timing based on real-time status data and combined with a user personalization model comprises dynamically adjusting the outbound calling timing based on real-time collected time scenario data and time period feature tags and answering feature tags in the user personalization model, including: Based on real-time collected time scenario data, determine whether the current time is weekday morning, weekday noon, weekday evening, weekday night, or non-weekday morning, non-weekday noon, non-weekday evening, or non-weekday night, and obtain the corresponding work and rest stability coefficient for that time period; The historical call answering frequency and the current time period rejection rate are obtained based on the time period feature tags and call answering feature tags in the current user personalization model. Based on historical call frequency, current time period rejection rate, and corresponding work-rest stability coefficient, a weighted sum is used to quantify the current time period's suitability score. When the current time period's suitability score is greater than or equal to a threshold, the current time period is considered a preferred call window. When the current time period's suitability score is less than the threshold, the current time period is considered a prohibited outbound call window. When the current time period is a prohibited outbound call window, the outbound call task is suspended. At the same time, the current outbound call task is cached and delayed until the user's preferred call window for execution.

[0012] Optionally, in the sixth implementation of the first aspect of the present invention, the step of combining the adjusted outbound calling method and outbound calling sequence with AI voice scripts adapted to user personalization characteristics and the current scenario to generate an intelligent communication strategy for the current order and initiate targeted delivery communication to the corresponding user includes: Based on real-time collected dynamic data of package delivery and environmental scene data, the current delivery scene is identified as either a Level 1 emergency scene or a Level 2 regular scene, and the corresponding basic script template is matched. Read the delivery method preference features in the user's personalized model. If the delivery method preference is collection on delivery, retrieve the name, address, and service area of ​​the collection point with the highest probability of use for the user, and generate a basic script template for the collection point. If the delivery method preference is door-to-door delivery, select the basic script for door-to-door appointment. The complaint risk constraints are generated by reading the complaint feature tags in the user's personalized model; wherein, the complaint risk constraints include: if there is an outbound call harassment complaint tag, simplify the wording length and limit the duration of a single call; if there is a wording dissatisfaction complaint tag, replace the standardized wording with a softer and more concise wording. The AI ​​voice scripts are adapted to the user's personalized characteristics and the current scenario by integrating basic script templates for various scenarios, such as those for collection points / door-to-door appointments, and complaint risk control constraints. The dynamically adjusted outbound calling method and timing are integrated with the AI ​​voice script to generate an intelligent communication strategy for the current order. Based on the intelligent communication strategy, targeted communication is initiated to the corresponding user.

[0013] A second aspect of the present invention provides a pre-departure intelligent communication strategy optimization device for express delivery, characterized in that it comprises: The user personalization model building module is used to collect all historical logistics interaction data of users and build a user personalization model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics through big data tagging processing. The real-time status data acquisition module is used to collect dynamic data of package delivery, time scene data, and environmental scene data in real time to obtain real-time status data. The dynamic adjustment module is used to dynamically adjust the outbound calling method and timing based on real-time status data and user personalization models. The intelligent communication module is used to combine the adjusted outbound calling method and timing with AI voice scripts adapted to user personalization and the current scenario to generate an intelligent communication strategy for the current order and initiate targeted delivery communication to the corresponding user.

[0014] Optionally, in a first implementation of the second aspect of the present invention, the user personalization model construction module includes: The full data collection submodule is used to collect all historical logistics interaction data of users, including: historical parcel signing records, AI voice answering / rejection records, missed call feedback, human customer service complaints and suggestions records, historical collection point selection records, communication method preference records, and time period answering pattern data. The data preprocessing submodule is used to perform cleaning, deduplication and normalization on the collected full volume of user historical logistics interaction data to obtain the processed full volume of user historical logistics interaction data. The tagging submodule is used to perform big data tagging on the pre-processed full data of user historical logistics interactions, generating any one or more user tags, including call answering feature tags, time period feature tags, communication preference feature tags, delivery method preference feature tags, and complaint feature tags. The call answering feature tags include: historical call answering frequency; The time-period feature tags include: rejection rates for morning, noon, evening, and night. The communication preference feature tags include: voice preference, SMS preference, and mixed communication preference; The delivery method preference feature tags include: pick-up preference and door-to-door delivery preference; The complaint feature tags include: outbound call harassment complaints and complaints of dissatisfaction with the script; A personalized user model is constructed based on the included call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics.

[0015] Optionally, in a second implementation of the second aspect of the present invention, the system further includes: a collection point information acquisition module, used to retrieve the collection point information with the highest usage probability currently associated with the user when the delivery method preference feature label in the user's personalized model is collection preference.

[0016] Optionally, in a third implementation of the second aspect of the present invention, the real-time status data acquisition module includes: The dynamic data of package delivery includes: package category identification, remaining delivery time, overdue warning level, delivery delay duration, real-time distance between the delivery person and the delivery address, package status, and package abnormality mark. The time-based scenario data includes: weekday morning, noon, evening, and nighttime periods, and non-weekday morning, noon, evening, and nighttime periods; The environmental scene data includes: weather information and road condition information.

[0017] Optionally, in a fourth implementation of the second aspect of the present invention, the dynamic adjustment module dynamically adjusts the outbound calling method based on real-time collected dynamic data of package delivery and environmental scene data, combined with communication preference feature tags in the user's personalized model, including: The scene recognition submodule is used to identify whether the current scene is a Level 1 emergency scene or a Level 2 normal scene based on real-time collected dynamic data of package delivery and environmental scene data. The outbound call mode adjustment submodule is used to dynamically adjust to voice-priority outbound calls when the current scenario is a level 1 emergency scenario and the communication preference feature label in the user's personalized model is the SMS preference feature label or the mixed communication preference feature label. When the current scenario is a level 2 routine scenario, the corresponding communication preference feature labels in the user's personalized model are maintained; The first-level emergency scenarios include any one or more of the following: package category identification prioritizes delivery categories and non-receiveable categories, time-limited signature packages, remaining delivery time is less than a threshold, overtime warning level meets preset requirements, delivery delay, real-time distance between delivery personnel and delivery address is less than a threshold, package status is pending delivery, package is abnormal, weather information is extreme rain or snow that meets preset requirements, or road condition information is road closure that meets preset requirements. The secondary routine scenarios include scenarios other than the primary emergency scenarios.

[0018] Optionally, in a fifth implementation of the second aspect of the present invention, the dynamic adjustment module dynamically adjusts the outbound call sequence based on real-time collected time scenario data and time period feature tags and answering feature tags in the user's personalized model, including: The work-rest stability coefficient acquisition submodule is used to determine whether the current time is a weekday morning, weekday noon, weekday evening, or weekday night, or a non-weekday morning, non-weekday noon, non-weekday evening, or non-weekday night, based on real-time collected time scenario data, and to obtain the work-rest stability coefficient corresponding to that time period. The outbound call window judgment submodule is used to obtain the historical answer frequency and the current time period rejection rate based on the time period feature tags and answer feature tags in the current user's personalized model, respectively. Based on the historical answer frequency, the current time period rejection rate, and the corresponding work-rest stability coefficient of the current time period, the current time period adaptation score is quantified by weighted summation. When the current time period adaptation score is greater than or equal to the threshold, the current time period is considered a preferred answer window. When the current time period adaptation score is less than the threshold, the current time period is considered a prohibited outbound call window. When the current time period is a prohibited outbound call window, the outbound call task is suspended. At the same time, the current outbound call task is cached and delayed until the user's preferred answer window for execution.

[0019] Optionally, in a sixth implementation of the second aspect of the present invention, the intelligent communication module includes: Based on real-time collected dynamic data of package delivery and environmental scene data, the current delivery scene is identified as either a Level 1 emergency scene or a Level 2 regular scene, and the corresponding basic script template is matched. Read the delivery method preference features in the user's personalized model. If the delivery method preference is collection on delivery, retrieve the name, address, and service area of ​​the collection point with the highest probability of use for the user, and generate a basic script template for the collection point. If the delivery method preference is door-to-door delivery, select the basic script for door-to-door appointment. The complaint risk constraints are generated by reading the complaint feature tags in the user's personalized model; wherein, the complaint risk constraints include: if there is an outbound call harassment complaint tag, simplify the wording length and limit the duration of a single call; if there is a wording dissatisfaction complaint tag, replace the standardized wording with a softer and more concise wording. The AI ​​voice scripts are adapted to the user's personalized characteristics and the current scenario by integrating basic script templates for scenarios, basic script templates for collection points / basic scripts for door-to-door appointments, and complaint risk control constraints. The dynamically adjusted outbound calling method and timing are integrated with the AI ​​voice script to generate an intelligent communication strategy for the current order. Based on the intelligent communication strategy, targeted communication is initiated to the corresponding user.

[0020] A third aspect of the present invention provides an electronic device, the electronic device comprising a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the electronic device to execute the various steps of the above-described intelligent communication strategy optimization method for pre-delivery express delivery.

[0021] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the various steps of the pre-delivery intelligent communication strategy optimization method for express delivery as described above.

[0022] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention integrates four steps: user historical behavior modeling, real-time multi-source data collection, dynamic adjustment of outbound call methods and timing, and personalized intelligent communication. It connects the data linkage and analysis links of the user side, order side, and environment side, abandons the traditional batch homogeneous outbound call mode, realizes personalized communication for each person and differentiated scheduling for each order, and coordinates and balances the core needs of last-mile delivery fulfillment efficiency and user non-disturbing service. It is suitable for large-scale express delivery outlets to carry out 24 / 7 pre-departure communication operations. 2. For urgent orders such as those with tight delivery times, package abnormalities, or extreme road conditions, this invention can bypass the user's SMS / mixed preferences to force voice-priority outbound calls, ensuring that urgent delivery information is delivered in a timely manner; for regular orders, it fully complies with the user's native communication preferences, taking into account both the risk control of order delay compensation and the user's right to choose their own communication. 3. This invention quantitatively assesses users' willingness to answer calls, initiates outbound calls only during periods of high user engagement, and suspends and caches outbound call tasks during periods of low willingness. At the same time, it sets a red line for prohibiting outbound calls at night, which significantly reduces the frequency of invalid outbound calls, improves the connection rate of pre-dispatch communication, and avoids the problem of disturbing the public by outbound calls from a temporal perspective. 4. This invention provides differentiated traffic limiting and script rewriting for two types of complaints: harassment complaints and dissatisfaction with the script. It also adapts the communication content to two delivery scenarios: collection and door-to-door delivery. This allows for customized AI communication scripts for each order, controlling communication details from all dimensions, including script duration, tone, and content. This reduces secondary complaints caused by inappropriate scripts and high-frequency calls from the source. Attached Figure Description

[0023] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a first flowchart of the intelligent communication strategy optimization method for pre-delivery of express delivery provided in an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of a pre-delivery intelligent communication strategy optimization device provided in an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0026] This invention provides a method, apparatus, device, and medium for optimizing intelligent communication strategies before express delivery. The method includes: collecting all historical logistics interaction data of users; constructing a personalized user model through big data tagging processing, encompassing answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics; collecting real-time dynamic data of package delivery, time scenario data, and environmental scenario data to obtain real-time status data; dynamically adjusting the outbound calling method and timing based on the real-time status data and the personalized user model; and generating an intelligent communication strategy for the current order by combining the adjusted outbound calling method and timing with AI voice scripts adapted to the user's personalized characteristics and the current scenario, and initiating targeted delivery communication to the corresponding user. This invention solves the technical problems of existing technologies that rely solely on simple, fixed program logic to perform outbound calling operations, lacking in-depth mining of user personalized characteristics, fusion analysis of real-time scenario data, and dynamic strategy adaptive adjustment capabilities. These problems have long resulted in a large volume of invalid outbound calls, wasted communication resources, low user connection rates, high complaint rates, and low delivery communication efficiency.

[0027] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the pre-delivery intelligent communication strategy optimization method for express delivery in this invention includes: 101. Collect all historical logistics interaction data of users, and construct a personalized user model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics through big data tagging processing. In this embodiment, the collected user's historical logistics interaction data includes: user's historical package receipt records, AI voice call answering / rejection records, missed call feedback, human customer service complaint and suggestion records, historical collection point selection records, communication method preference records, and time period call pattern data; The collected full data of user historical logistics interactions is processed including data cleaning, data deduplication and normalization to obtain the processed full data of user historical logistics interactions. The data cleaning process includes: removing invalid calls with a duration of less than 3 seconds, outbound calls automatically tested by the system, user-suspended or disconnected numbers, and user interaction data when the user turns off their phone. Such data does not reflect the user's subjective interaction intention, and removing it can eliminate the interference of invalid data on the tags and ensure the objectivity of the profile. The data deduplication includes: merging repeated outbound call data for the same order on a single day into one interaction data record, retaining only the first outbound call interaction result, avoiding the platform's batch retrying of outbound calls causing duplicate accumulation of interaction data, and preventing user call frequency and rejection rate tag scores from being distorted and too high; The normalization process includes: unifying all call answering, time period, and preference-related feature data into a score range of 0-1, eliminating differences in duration, frequency, and percentage units, and improving the model's computation speed and judgment accuracy.

[0029] Based on the pre-processed full volume of user historical logistics interaction data, big data tagging is performed to generate any one or more user tags, including call answering feature tags, time period feature tags, communication preference feature tags, delivery method preference feature tags, and complaint feature tags. The call answering feature tags include: historical call answering frequency; The time-period feature tags include: rejection rates for morning, noon, evening, and night. The communication preference feature tags include: voice preference, SMS preference, and mixed communication preference; among them, voice preference includes: historical voice connection interaction rate ≥65%, users are willing to communicate delivery matters in real time via voice; SMS preference includes: actively replying to SMS, blocking AI voice outbound calls, historical voice rejection rate exceeding 80%, resisting voice call disturbances; mixed communication preference includes: no obvious single channel preference, response time and response probability of both types of channels are equal, and the system can adaptively schedule channels.

[0030] The delivery method preference tags include: collection on delivery preference and door-to-door delivery preference; specifically, a collection on delivery preference is determined if the proportion of collection on delivery within 90 days is ≥60%, indicating that users prefer self-service pickup at delivery stations and avoid waiting at the door; a door-to-door delivery preference is determined if the proportion of door-to-door signing within 90 days is ≥60%, indicating that users have a strong need for in-person signing and verification; a difference of less than 10% indicates a two-way matching preference, where two delivery options are provided simultaneously during communication for users to choose from.

[0031] When the delivery method preference feature label in the user's personalized model is collection preference, retrieve the collection point information with the highest probability of use associated with the current user.

[0032] The complaint feature tags include: outbound call harassment complaints and complaints of dissatisfaction with the wording; specifically, if there are 2 or more complaints of pre-deployment outbound call harassment in the past 90 days, the user will be marked with the outbound call harassment complaint tag and included in the high-frequency risk control user pool; if there are 2 or more complaints of stiff wording, verbose wording, or harsh tone in the past 90 days, the user will be marked with the wording dissatisfaction complaint tag and bound to exclusive wording risk control rules. Both types of tags will be retained indefinitely until the user has no similar complaints for 90 consecutive days and the tags will be automatically removed.

[0033] A personalized user model is constructed based on the included call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics.

[0034] 102. Collect real-time dynamic data of package delivery, time scene data, and environmental scene data to obtain real-time status data; In this embodiment, the dynamic data of package delivery includes: package category identifier, remaining delivery time, overdue warning level, delivery delay duration, real-time distance between the delivery person and the delivery address, package retention status, and package abnormality mark. The package categories are labeled as follows: fresh and cold chain products, medicines with time limits, documents and certificates with time limits, general daily necessities, large and heavy items, and prohibited items for collection points. The categories are linked to the delivery priority rules, with fresh and medicines being given the highest priority, while large and prohibited items are directly removed from the collection options, thus avoiding delivery disputes from the source.

[0035] Timeout warning levels: Level 1 Red Warning, Level 2 Yellow Warning, Level 3 Green Normal; the level is linked to the platform's timeout compensation rules. Orders with a red warning are highly likely to trigger platform compensation, and the user must be contacted to confirm receipt in advance.

[0036] Delivery delay duration: The difference between the delivery time and the standard delivery time planned by the logistics system. It excludes human delays caused by users actively changing the delivery time and only counts objective delays caused by the road network, network points and transportation capacity to ensure fair and accurate scenario judgment.

[0037] Package retention status includes five categories: sorting retention, network point retention, delivery in progress retention, waiting for door-to-door delivery, and assigned delivery person; Package anomaly markers: damaged, incomplete address, cold chain loss anomaly markers; The time-based scenario data includes: weekday morning, noon, evening, and nighttime periods, and non-weekday morning, noon, evening, and nighttime periods; The environmental scene data includes: weather information and road condition information.

[0038] 103. Based on real-time status data and combined with user personalization models, dynamically adjust outbound calling methods and timing. In this embodiment, based on real-time collected dynamic data on package delivery and environmental scenarios, combined with communication preference feature tags in the user's personalized model, the outbound calling method is dynamically adjusted, including: Based on real-time collected dynamic data on package delivery and environmental scene data, the current scenario is identified as either a Level 1 emergency scenario or a Level 2 routine scenario. When the current scenario is a Level 1 emergency scenario, and the communication preference feature label in the user's personalized model is the SMS preference feature label or the mixed communication preference feature label, then the call will be dynamically adjusted to prioritize voice calls. When the current scenario is a level 2 routine scenario, the corresponding communication preference feature labels in the user's personalized model are maintained; The first-level emergency scenarios include any one or more of the following: packages identified as priority delivery categories and non-receiveable categories, packages with time-limited signing, remaining delivery time less than a threshold, overtime warning level meeting preset requirements, delivery delays, real-time distance between the delivery person and the delivery address less than a threshold, packages in a pending delivery status, packages with abnormalities, extreme rain or snow meeting preset requirements in weather information, or road closures meeting preset requirements in road condition information. The secondary routine scenarios include scenarios other than the primary emergency scenarios.

[0039] The second-level routine scenarios include ordinary daily necessities delivery, orders with sufficient timeliness, normal weather and road conditions, and no abnormalities in the package. These orders have no risk of delay, loss, or failure to fulfill obligations. It is sufficient to respect the user's independent communication preferences and there is no need to force urgent voice calls to minimize user disturbance.

[0040] Based on real-time collected time-scene data and time-slot feature tags and call answering feature tags in the user's personalized model, the outbound call timing is dynamically adjusted, including: Based on real-time collected time scenario data, determine whether the current time is weekday morning, weekday noon, weekday evening, weekday night, or non-weekday morning, non-weekday noon, non-weekday evening, or non-weekday night, and obtain the corresponding work and rest stability coefficient for that time period; The historical call answering frequency and the current time period rejection rate are obtained based on the time period feature tags and call answering feature tags in the current user personalization model. Based on historical call frequency, current time period rejection rate, and corresponding work-rest stability coefficient, a weighted sum is used to quantify the current time period's suitability score. When the current time period's suitability score is greater than or equal to a threshold, the current time period is considered a preferred call window. When the current time period's suitability score is less than the threshold, the current time period is considered a prohibited outbound call window. When the current time period is a prohibited outbound call window, the outbound call task is suspended. At the same time, the current outbound call task is cached and delayed until the user's preferred call window for execution.

[0041] For example: Time period fit score S = A × 0.4 + (1 - B) × 0.4 + C × 0.2; Where A = historical call frequency normalized score, with high-frequency calls approaching 1 and low-frequency calls approaching 0; B = current user active rejection rate, where (1-B) is the user's willingness to answer calls score for the current time period, with a higher rejection rate and higher willingness to answer; C = time period rest stability coefficient. The three dimensions are weighted and coupled to comprehensively quantify the suitability of outbound calls during different time periods, avoiding the one-sidedness of judging time periods based on single data.

[0042] In this embodiment, when the matching score S ≥ 0.6, a preferred answering window is selected, allowing normal outbound calls to be initiated; at this score, the user answering probability exceeds 60%, resulting in the optimal outbound call return on investment. When the matching score S < 0.6, an outbound call window is prohibited, and the current outbound call task is suspended in real time; at this score, the user's willingness to answer is extremely low, the outbound call is likely to be invalid and easily causes resentment, so it is directly suspended to avoid invalid calls. When the matching score S < 0.6 for the current time period, the next preferred window with a matching score ≥ 0.6 for the day is automatically matched to execute the outbound call, automatically sorting the optimal answering time period, without the need for manual scheduling, and with fully automatic delayed scheduling. Nighttime full coverage: From 21:00 to 7:00 the next day, an outbound call window is prohibited across the entire platform, and all users suspend voice outbound calls. This is a platform compliance-compliant do-not-disturb rule and is not affected by order urgency level, user preferences, or matching score.

[0043] 104. By combining the adjusted outbound calling methods and timing with AI voice scripts adapted to user personalization and the current scenario, generate an intelligent communication strategy for the current order and initiate targeted delivery communication to the corresponding user.

[0044] In this embodiment, based on real-time collected dynamic data of package delivery and environmental scene data, the current delivery scenario is identified as a Level 1 emergency scenario or a Level 2 routine scenario, and the corresponding scenario basic script template is matched. Read the delivery method preference features in the user's personalized model. If the delivery method preference is collection on delivery, retrieve the name, address, and service area of ​​the collection point with the highest probability of use for the user, and generate a basic script template for the collection point. If the delivery method preference is door-to-door delivery, select the basic script for door-to-door appointment. The complaint risk constraints are generated by reading the complaint feature tags in the user's personalized model; wherein, the complaint risk constraints include: if there is an outbound call harassment complaint tag, simplify the script length and limit the duration of a single call; if there is a script dissatisfaction complaint tag, replace the standardized script with a soft and concise script. The AI ​​voice scripts are adapted to the user's personalized characteristics and the current scenario by integrating basic script templates for scenarios, basic script templates for collection points / basic scripts for door-to-door appointments, and complaint risk control constraints. The dynamically adjusted outbound calling method and timing are integrated with the AI ​​voice script to generate an intelligent communication strategy for the current order. Based on the intelligent communication strategy, targeted communication is initiated to the corresponding user.

[0045] Compared to the traditional batch AI outbound calling mode with unified time periods, unified scripts, and fixed channels, this embodiment relies on a multi-dimensional user profile and a real-time scenario dual-linkage scheduling mechanism to achieve full-domain optimization from user adaptation, scenario adaptation, and risk control compliance. It accurately matches users' answering schedules and communication habits, avoids invalid disturbances and user complaints caused by inappropriate scripts, and maximizes the protection of user service experience.

[0046] The above describes the method for optimizing pre-delivery intelligent communication strategies in embodiments of the present invention. The following describes the device for optimizing pre-delivery intelligent communication strategies in embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the pre-departure intelligent communication strategy optimization device for express delivery includes: The user personalization model building module 201 is used to collect all historical logistics interaction data of users and build a user personalization model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics through big data tagging processing. In this embodiment, the user personalization model construction module 201 includes: The full data collection submodule 2011 is used to collect full data of users' historical logistics interactions, including: users' historical package signing records, AI voice answering / rejecting records, feedback from missed calls, records of complaints and suggestions from human customer service, historical collection point selection records, communication method preference records, and data on call patterns during different time periods. The data preprocessing submodule 2012 is used to perform cleaning, deduplication and normalization on the collected full volume of user historical logistics interaction data to obtain the processed full volume of user historical logistics interaction data. The tagging processing submodule 2013 is used to perform big data tagging processing on the pre-processed full volume of user historical logistics interaction data, generating any one or more user tags, including call answering feature tags, time period feature tags, communication preference feature tags, delivery method preference feature tags, and complaint feature tags. The call answering feature tags include: historical call answering frequency; The time-period feature tags include: rejection rates for morning, noon, evening, and night. The communication preference feature tags include: voice preference, SMS preference, and mixed communication preference; The delivery method preference feature tags include: collection on delivery preference and door-to-door delivery preference; when the delivery method preference feature tag in the user's personalization model is collection on delivery preference, the collection point information with the highest usage probability associated with the current user is retrieved.

[0047] The complaint feature tags include: outbound call harassment complaints and complaints of dissatisfaction with the script; A personalized user model is constructed based on the included call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics.

[0048] The real-time status data acquisition module 202 is used to collect real-time dynamic data of package delivery, time scene data, and environmental scene data to obtain real-time status data. In this embodiment, the real-time status data acquisition module 202 includes: The dynamic data of package delivery includes: package category identification, remaining delivery time, overdue warning level, delivery delay duration, real-time distance between the delivery person and the delivery address, package status, and package abnormality mark. The time-based scenario data includes: weekday morning, noon, evening, and nighttime periods, and non-weekday morning, noon, evening, and nighttime periods; The environmental scene data includes: weather information and road condition information.

[0049] The dynamic adjustment module 203 is used to dynamically adjust the outbound calling method and outbound calling sequence based on real-time status data and user personalization model; In this embodiment, the dynamic scheduling module 203 includes: The outbound call method adjustment submodule 2031 is used to dynamically adjust the outbound call method based on real-time collected dynamic data of package delivery and environmental scenario data, combined with communication preference feature tags in the user's personalized model. In this embodiment, the outbound call mode adjustment submodule 2031 includes: The scene recognition submodule 20311 is used to identify whether the current scene is a level one emergency scene or a level two normal scene based on the real-time collected dynamic data of package delivery and environmental scene data. The outbound call mode adjustment submodule 20312 is used to dynamically adjust to voice-priority outbound calls when the current scenario is a level one emergency scenario and the communication preference feature label in the user's personalized model is an SMS preference feature label or a mixed communication preference feature label. When the current scenario is a level 2 routine scenario, the corresponding communication preference feature labels in the user's personalized model are maintained; The first-level emergency scenarios include any one or more of the following: package category identification prioritizes delivery categories and non-receiveable categories, time-limited signature packages, remaining delivery time is less than a threshold, overtime warning level meets preset requirements, delivery delay, real-time distance between delivery personnel and delivery address is less than a threshold, package status is pending delivery, package is abnormal, weather information is extreme rain or snow that meets preset requirements, or road condition information is road closure that meets preset requirements. The secondary routine scenarios include scenarios other than the primary emergency scenarios.

[0050] This embodiment achieves a two-way balance between order fulfillment risk and user experience by forcibly switching to voice outbound calls in emergency scenarios and following user preferences in normal scenarios, while also taking into account compliance and fulfillment efficiency.

[0051] The outbound call timing adjustment submodule 2032 dynamically adjusts the outbound call timing based on real-time collected time scenario data and time period feature tags and answering feature tags in the user's personalized model. In this embodiment, the outbound call timing adjustment submodule 2032 includes: The work-rest stability coefficient acquisition submodule 20321 is used to determine whether the current time is a weekday morning, weekday noon, weekday evening, or weekday night, or a non-weekday morning, non-weekday noon, non-weekday evening, or non-weekday night, based on the real-time collected time scenario data, and to obtain the work-rest stability coefficient corresponding to that time period. The outbound call window judgment submodule 20322 is used to obtain the historical answer frequency and the current time period rejection rate based on the time period feature tags and answer feature tags in the current user's personalized model, respectively; based on the historical answer frequency, the current time period rejection rate, and the corresponding work-rest stability coefficient of the current time period, the current time period adaptation score is quantified by weighted summation; when the current time period adaptation score is greater than or equal to the threshold, the current time period is considered a preferred answer window; when the current time period adaptation score is less than the threshold, the current time period is considered a prohibited outbound call window; when the current time period is a prohibited outbound call window, the outbound call task is suspended; at the same time, the current outbound call task is cached and delayed until the user's preferred answer window for execution.

[0052] This embodiment uses a three-dimensional weighted matching score based on willingness to answer calls, rejection rate, and work-rest coefficient to avoid the one-sidedness of judging time periods with a single data point and improve the accuracy of outbound call window determination. By using a fixed matching score to divide the threshold features of preferred / prohibited windows, it achieves optimal outbound calls, ensures the best outbound call return on investment, and improves overall connection efficiency.

[0053] The intelligent communication module 204 is used to combine the adjusted outbound calling method and outbound calling sequence with AI voice scripts adapted to the user's personalized characteristics and the current scenario to generate an intelligent communication strategy for the current order and initiate targeted delivery communication to the corresponding user.

[0054] In this embodiment, the intelligent communication module 204 includes: Based on real-time collected dynamic data of package delivery and environmental scene data, the current delivery scene is identified as either a Level 1 emergency scene or a Level 2 regular scene, and the corresponding basic script template is matched. Read the delivery method preference features in the user's personalized model. If the delivery method preference is collection on delivery, retrieve the name, address, and service area of ​​the collection point with the highest probability of use for the user, and generate a basic script template for the collection point. If the delivery method preference is door-to-door delivery, select the basic script for door-to-door appointment. The complaint risk constraints are generated by reading the complaint feature tags in the user's personalized model; wherein, the complaint risk constraints include: if there is an outbound call harassment complaint tag, simplify the script length and limit the duration of a single call; if there is a script dissatisfaction complaint tag, replace the standardized script with a soft and concise script. The AI ​​voice scripts are adapted to the user's personalized characteristics and the current scenario by integrating basic script templates for scenarios, basic script templates for collection points / basic scripts for door-to-door appointments, and complaint risk control constraints. The dynamically adjusted outbound calling method and timing are integrated with the AI ​​voice script to generate an intelligent communication strategy for the current order. Based on the intelligent communication strategy, targeted communication is initiated to the corresponding user.

[0055] This embodiment achieves scenario-based speech matching by hierarchically adapting speech template features such as speaking speed, duration, and sentence structure for emergency / routine scenarios, thereby improving the efficiency of user information reception.

[0056] above Figure 2 The intelligent communication strategy optimization device for pre-delivery of express delivery in this embodiment of the invention will be described in detail from the perspective of modular functional entities. The electronic device in this embodiment of the invention will be described in detail from the perspective of hardware processing.

[0057] Figure 3This is a schematic diagram of the structure of an electronic device 700 provided in an embodiment of the present invention. The electronic device 700 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 710 (e.g., one or more processors) and a memory 720, and one or more storage media 730 (e.g., one or more mass storage devices) for storing application programs 733 or data 732. The memory 720 and storage media 730 can be temporary or persistent storage. The program stored in the storage media 730 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the electronic device 700. Furthermore, the processor 710 may be configured to communicate with the storage media 730 and execute the series of instruction operations in the storage media 730 on the electronic device 700.

[0058] Electronic device 700 may also include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input / output interfaces 750, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The illustrated electronic device structure does not constitute a limitation on electronic devices and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0059] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the pre-delivery intelligent communication strategy optimization method.

[0060] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0061] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0062] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing intelligent communication strategies before express delivery, characterized in that, include: Collect all historical logistics interaction data of users, and through big data tagging processing, construct a personalized user model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics; Real-time data collection of dynamic parcel delivery data, time-based data, and environmental data is used to obtain real-time status data. Based on real-time status data and combined with user-personalized models, the outbound calling method and timing are dynamically adjusted. By combining the adjusted outbound calling methods and timings with AI voice scripts tailored to user personalities and the current scenario, an intelligent communication strategy is generated for the current order, and targeted delivery communication is initiated to the corresponding user.

2. The method for optimizing intelligent communication strategies before express delivery as described in claim 1, characterized in that, The process involves collecting all historical logistics interaction data from users and, through big data tagging, constructing a personalized user model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics. The data collected includes: user's historical package receipt records, AI voice call answering / rejection records, missed call feedback, human customer service complaint and suggestion records, historical collection point selection records, communication method preference records, and time period call pattern data. The collected full volume of user historical logistics interaction data is processed by cleaning, deduplication and normalization to obtain the processed full volume of user historical logistics interaction data. Based on the pre-processed full volume of user historical logistics interaction data, big data tagging is performed to generate any one or more user tags, including call answering feature tags, time period feature tags, communication preference feature tags, delivery method preference feature tags, and complaint feature tags. The call answering feature tags include: historical call answering frequency; The time-period feature tags include: rejection rates for morning, noon, evening, and night. The communication preference feature tags include: voice preference, SMS preference, and mixed communication preference; The delivery method preference feature tags include: pick-up preference and door-to-door delivery preference; The complaint feature tags include: outbound call harassment complaints and complaints of dissatisfaction with the script; A personalized user model is constructed based on the included call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics.

3. The method for optimizing intelligent communication strategies before express delivery as described in claim 2, characterized in that, When the delivery method preference feature label in the user's personalized model is collection preference, retrieve the collection point information with the highest probability of use associated with the current user.

4. The method for optimizing intelligent communication strategies before express delivery as described in claim 1, characterized in that, The real-time collection of dynamic data on package delivery, time-based scenario data, and environmental scenario data includes: The dynamic data of package delivery includes: package category identification, remaining delivery time, overdue warning level, delivery delay duration, real-time distance between the delivery person and the delivery address, package status, and package abnormality mark. The time-based scenario data includes: weekday morning, noon, evening, and nighttime periods, and non-weekday morning, noon, evening, and nighttime periods; The environmental scene data includes: weather information and road condition information.

5. The method for optimizing intelligent communication strategies before express delivery as described in claim 1, characterized in that, The method of dynamically adjusting outbound calling methods and timing based on real-time status data and user personalization models involves dynamically adjusting outbound calling methods based on real-time collected dynamic data on package delivery and environmental scenarios, combined with communication preference feature tags in the user personalization model. This includes: Based on real-time collected dynamic data on package delivery and environmental scene data, the current scenario is identified as either a Level 1 emergency scenario or a Level 2 routine scenario. When the current scenario is a Level 1 emergency scenario, and the communication preference feature label in the user's personalized model is the SMS preference feature label or the mixed communication preference feature label, then the call will be dynamically adjusted to prioritize voice calls. When the current scenario is a level 2 routine scenario, the corresponding communication preference feature labels in the user's personalized model are maintained; The first-level emergency scenarios include any one or more of the following: packages identified as priority delivery categories and non-receiveable categories, packages with time-limited signing, remaining delivery time less than a threshold, overtime warning level meeting preset requirements, delivery delays, real-time distance between the delivery person and the delivery address less than a threshold, packages in a pending delivery status, packages with abnormalities, extreme rain or snow meeting preset requirements in weather information, or road closures meeting preset requirements in road condition information. The secondary routine scenarios include scenarios other than the primary emergency scenarios.

6. The method for optimizing intelligent communication strategies before express delivery as described in claim 1, characterized in that, The method of dynamically adjusting outbound calling methods and timing based on real-time status data and user personalization models involves dynamically adjusting the outbound calling timing according to real-time collected time scenario data and time period feature tags and answering feature tags in the user personalization model, including: Based on real-time collected time scenario data, determine whether the current time is weekday morning, weekday noon, weekday evening, weekday night, or non-weekday morning, non-weekday noon, non-weekday evening, or non-weekday night, and obtain the corresponding work and rest stability coefficient for that time period; The historical call answering frequency and the current time period rejection rate are obtained based on the time period feature tags and call answering feature tags in the current user personalization model. Based on historical call frequency, current time period rejection rate, and corresponding work-rest stability coefficient, a weighted sum is used to quantify the current time period's suitability score. When the current time period's suitability score is greater than or equal to a threshold, the current time period is considered a preferred call window. When the current time period's suitability score is less than the threshold, the current time period is considered a prohibited outbound call window. When the current time period is a prohibited outbound call window, the outbound call task is suspended. At the same time, the current outbound call task is cached and delayed until the user's preferred call window for execution.

7. The method for optimizing intelligent communication strategies before express delivery as described in claim 1, characterized in that, The process involves combining the adjusted outbound calling method and timing with AI voice prompts tailored to the user's personalized characteristics and the current scenario to generate an intelligent communication strategy for the current order, initiating targeted delivery communication to the corresponding user, including: Based on real-time collected dynamic data of package delivery and environmental scene data, the current delivery scene is identified as either a Level 1 emergency scene or a Level 2 regular scene, and the corresponding basic script template is matched. Read the delivery method preference features in the user's personalized model. If the delivery method preference is collection on delivery, retrieve the name, address, and service area of ​​the collection point with the highest probability of use for the user, and generate a basic script template for the collection point. If the delivery method preference is door-to-door delivery, select the basic script for door-to-door appointment. The complaint risk constraints are generated by reading the complaint feature tags in the user's personalized model; wherein, the complaint risk constraints include: if there is an outbound call harassment complaint tag, simplify the wording length and limit the duration of a single call; if there is a wording dissatisfaction complaint tag, replace the standardized wording with a softer and more concise wording. The AI ​​voice scripts are adapted to the user's personalized characteristics and the current scenario by integrating basic script templates for scenarios, basic script templates for collection points / basic scripts for door-to-door appointments, and complaint risk control constraints. The dynamically adjusted outbound calling method and timing are integrated with the AI ​​voice script to generate an intelligent communication strategy for the current order. Based on the intelligent communication strategy, targeted communication is initiated to the corresponding user.

8. A pre-delivery intelligent communication strategy optimization device for express delivery, characterized in that, include: The user personalization model building module is used to collect all historical logistics interaction data of users and build a user personalization model that includes call answering characteristics, time period characteristics, communication preference characteristics, delivery method preference characteristics, and complaint characteristics through big data tagging processing. The real-time status data acquisition module is used to collect dynamic data of package delivery, time scene data, and environmental scene data in real time to obtain real-time status data; The dynamic adjustment module is used to dynamically adjust the outbound calling method and timing based on real-time status data and user personalization models. The intelligent communication module is used to combine the adjusted outbound calling method and timing with AI voice scripts adapted to user personalization and the current scenario to generate an intelligent communication strategy for the current order and initiate targeted delivery communication to the corresponding user.

9. An electronic device comprising a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the pre-delivery intelligent communication strategy optimization method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement each step of the pre-delivery intelligent communication strategy optimization method for express delivery as described in any one of claims 1-7.

Images