Intelligent paperless office management method and system based on Internet of Things

Abstract

The invention relates to the technical field of office management, in particular to an intelligent paperless office management method and system based on the Internet of Things. An intelligent paperless office management method based on the Internet of Things comprises the following steps that S1, environment data related to a conference and conference agenda information preset before the conference are obtained through an Internet of Things sensor deployed in a conference room; s2, on the basis of the conference agenda information, constructing a concern and role linkage knowledge graph in combination with historical conference data, planning a role switching sequence of each participant at different agenda nodes on the basis of the knowledge graph before the conference is started, and generating a preset planning role switching path. By adopting a role switching mechanism combining preset planning and dynamic triggering, the roles of the participants can be automatically or manually switched and the corresponding note templates and permissions are issued according to the conference agenda and the real-time interaction condition, so that the problems of disordered role division and dispersed note information in the traditional conference are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of office management, and particularly relates to an intelligent paperless office management method and system based on Internet of Things. BACKGROUND

[0002] In modern enterprises, meetings have become the core link of information sharing, decision making and team collaboration. With the popularization of mobile Internet and Internet of Things technology, various intelligent terminals, large screen displays and collaborative whiteboard devices are widely used in meeting scenarios. However, traditional meetings still rely on paper meeting minutes, manual distribution of documents and simple and extensive role division, resulting in problems such as untimely information transmission, unclear role responsibilities and low efficiency of on-site collaboration.

[0003] With the increasing demand for cross-regional and heterogeneous team collaboration, the scale and complexity of meetings are constantly rising. Manual distribution of materials and oral division of labor can no longer meet the collaboration needs. In existing paperless meetings, personnel are divided into different roles, and static permission allocation is performed. However, this method cannot meet the real-time change of personnel permissions. For example, when a technical expert needs to become a speaker, it is found that the editing permission cannot be quickly adjusted. Moreover, when the role is switched, the permission may be confused, for example, the permission of the host appears on the listener, and the risk of sensitive information leakage increases due to lax permission management. SUMMARY

[0004] In order to overcome the shortcomings of being unable to flexibly schedule roles and permissions, the present application provides an intelligent paperless office management method and system based on Internet of Things.

[0005] Technical scheme: An intelligent paperless office management method based on Internet of Things, comprising the following steps:

[0006] S1: acquiring conference-related environmental data and pre-set conference agenda information through an Internet of Things sensor deployed in a conference room;

[0007] S2: based on the conference agenda information, constructing a concern point and role linkage knowledge graph in combination with historical conference data, and planning the role switching order of each participant at different agenda nodes based on the knowledge graph before the start of the conference, to generate a pre-set planned role switching path;

[0008] S3: during the conference, automatically or manually switching the roles of the participants according to the pre-set planned role switching path, and assigning corresponding note templates and permissions to the roles, and monitoring and adjusting the temporary planned role switching path;

[0009] S4: express the permission set after each role switching as a fine-grained permission vector, calculate the Mahalanobis distance based on the permission vector and the expected permission vector in the historical normal state, monitor the abnormal points and make adjustments;

[0010] S5: use the permission recovery formula to recover the permissions when the abnormal state is removed;

[0011] S6: synchronize the modifications made by each participant to the conference files in real time to the conference server, and save the modification results of each participant in the conference storage module in a versioned manner after the conference ends; the versioned saving should include the modifier identification, modification timestamp and change summary.

[0012] Preferably, the conference-related environmental data and the pre-conference preset conference agenda information are obtained through the Internet of Things sensors deployed in the conference room, including: the conference-related environmental data includes the noise level, temperature, humidity, light intensity, air quality in the conference room, and the personnel distribution collected by the seat sensor; the pre-conference preset conference agenda information includes the content theme, stage time arrangement and required leading role type of each agenda node.

[0013] Preferably, based on the conference agenda information, a focus and role linkage knowledge graph is constructed in combination with historical conference data, and the role switching order of each participant at different agenda nodes is planned based on the knowledge graph before the conference starts, and a preset planned role switching path is generated, including: labeling the content of the participants' focus on the documents in the historical conference to record triples, the triples include conference file identification, focus label and role state, based on the triples of a preset number of historical conferences, a focus and role linkage knowledge graph is constructed, according to the focus and role linkage knowledge graph, an optimal preset planned role switching path before the conference is constructed, according to the focus and role linkage knowledge graph and the conference-related environmental data, a focus deviation is detected during the conference, and a temporary path branch is generated, wherein the focus deviation is the degree to which the current participant's focus on the agenda theme deviates from the historical benchmark; the focus and role linkage knowledge graph is stored in a graph database, the nodes include file identification nodes, focus label nodes, role state nodes and switching path nodes, and the edges represent the multi-dimensional association between the nodes; and a graph mining algorithm is used to identify high-frequency focus combinations and typical conversion modes; the temporary path branch is a local insertion of a role switching sub-process at a node of the preset planned role switching path, which only supplements, deletes or adjusts the order of the local theme or hot spot.

[0014] Preferably, the preset planning role switching path is constructed according to the concern-point-role linkage knowledge graph before the meeting, the concern deviation is detected during the meeting according to the concern-point-role linkage knowledge graph, and a temporary path branch is generated, including: intelligently recommending an optimal preset planning path based on the concern tag distribution of similar historical meetings before the meeting according to the concern-point-role linkage knowledge graph; detecting the deviation of the current concern tag from the historical distribution during the meeting, and automatically prompting or inserting a temporary path branch; forming a cross-session concern flow record after the meeting; and setting a timeout rollback mechanism, which is to cancel all associated operations after role switching.

[0015] Preferably, during the meeting, the roles of the participants are automatically or manually switched according to the preset planning role switching path, and the corresponding note templates and permissions are assigned to the roles, and the temporary planning role switching path is monitored and adjusted, including: based on the collected conference environment data and the pre-set conference agenda, the roles of the participants are automatically or manually switched according to the conference process in the preset planning role switching path, and the corresponding note templates and permissions are assigned to each role, during the meeting, the speech content and interaction behavior of the participants are analyzed in real time, when the rule trigger of the temporary planning role switching path is detected, the temporary role switching instruction is dynamically generated and sent to the client, and the note tool and the permission corresponding to the switched role are switched; wherein the preset planning role switching path is the role switching sequence planned in advance for each participant at each agenda node before the meeting; the temporary planning role switching path is a complete role switching sequence that replaces the remaining part of the preset planning role switching path based on the real-time detected concern deviation or other trigger factors.

[0016] Preferably, the permission set after each role switching is represented as a fine-grained permission vector, the Mahalanobis distance is calculated based on the permission vector and the expected permission vector in the historical normal state, the abnormal points are monitored and adjusted, including: the permission set after each role switching is represented as a vector P i =[p i,1 ,p i,2 ……,p i,m ] T , wherein p i,j ∈{0,1} represents whether the i-th role switching has the j-th fine-grained permission; m is the total number of all fine-grained permissions defined in advance; the Mahalanobis distance square is obtained by using the Mahalanobis distance formula, and the smoothed statistics are obtained by performing EWMA smoothing on the Mahalanobis distance; when the Mahalanobis distance square is greater than or equal to a preset threshold, the current role switching is regarded as an abnormal point, and the transaction rollback or permission isolation is performed; when the abnormal state is removed, the permission is restored by using the permission recovery formula, wherein the Mahalanobis distance formula is:

[0017]

[0018] wherein, is the Mahalanobis distance square; P i is the fine-grained permission set vector held by the current user after the i-th role switching; μ i is the expected value of the permission vector in the historical normal switching process in the i-th role switching stage; α i is the covariance matrix of the fine-grained permission set vector in the i-th role switching stage;

[0019] denotes the inverse of the covariance matrix; () T is the transpose.

[0020] Preferably, the restoring the permissions using the permission restoring formula after the abnormal state is removed, comprises: wherein the permission restoring formula is:

[0021]

[0022] wherein, is the permission vector to be re-issued to the user after the abnormal detection in the i-th role switching stage; μ i is the expected value of the permission vector in the historical normal switching process in the i-th role switching stage; α ′ is the recovery coefficient based on EWMA smoothing; β is a temporary path influence weight, β ∈ [0, 1], and the larger the value is, the more the historical influence of the temporary branch is considered, wherein α ′ + β ≤ 1; ΔB i is the average deviation vector of the historical preset number of temporary branches and expectations, reflecting the contribution direction and strength of the temporary path to the permission allocation; P min is the minimum permission set, containing the lowest permission to guarantee the basic participation ability of the conference; α ′ is the recovery coefficient; z is a preset threshold; S i is a smoothing statistic.

[0023] Preferably, the average deviation vector of the historical preset number of temporary branches and expectations, comprises: obtaining the average deviation vector of the historical preset number of temporary branches and expectations through a temporary vector calculation formula, wherein the temporary vector calculation formula is:

[0024]

[0025] wherein, ΔB i is the average deviation vector of the historical preset number of temporary branches and expectations; K is the historical preset number; is the actual permission vector corresponding to the k-th temporary switching branch in the i-th role switching stage; μi The expected value of the authority vector in the historical normal switching process under the stage of the i-th role switching.

[0026] Preferably, the modifications made by each participant to the conference files during the conference are synchronized to the conference server in real time, and the modification results of each participant are saved in the conference storage module in a versioned manner after the conference ends; the versioned saving should include the modifier identification, modification timestamp and change summary, including: before the conference, the participants are arranged by time period according to the conference agenda and participant schedule, and the required files of the conference are pre-delivered and cached to each participant terminal; after the conference, the modification contents of each participant are summarized and saved in the conference storage module.

[0027] An intelligent paperless office management system based on the Internet of Things, comprising:

[0028] A data acquisition module is configured to acquire conference-related environmental data and pre-set conference agenda information through Internet of Things sensors deployed in the conference room.

[0029] A path construction module is configured to construct a concern point and role linkage knowledge graph based on the conference agenda information and historical conference data, and plan the role switching sequence of each participant at different agenda nodes based on the knowledge graph before the conference starts, to generate a pre-set planned role switching path.

[0030] A monitoring module switching module is configured to automatically or manually switch the participant roles according to the pre-set planned role switching path during the conference, and assign corresponding note templates and permissions to the roles, and monitor and adjust the temporary planned role switching path.

[0031] An abnormality monitoring module is configured to represent the permission set after each role switching as a fine-grained permission vector, and perform Mahalanobis distance calculation based on the permission vector and the expected permission vector in the historical normal state to monitor abnormal points and make adjustments.

[0032] A permission recovery module is configured to recover the permissions using a permission recovery formula when the abnormal state is resolved.

[0033] A data storage module is configured to synchronize the modifications made by each participant to the conference files during the conference to the conference server in real time, and save the modification results of each participant in the conference storage module in a versioned manner after the conference ends; the versioned saving should include the modifier identification, modification timestamp and change summary.

[0034] Compared with the prior art, the present application has the following advantages:

[0035] 1、 The application can automatically or manually switch the roles of participants and issue corresponding note templates and permissions according to the meeting agenda and real-time interaction, avoid the problems of role confusion and scattered note information in traditional meetings, by adopting a role switching mechanism combining preset planning and dynamic triggering; based on the attention point and role linkage knowledge graph, the optimal preset switching path is intelligently recommended, the preparation workload before the meeting is reduced, and the entire meeting process is smoother;

[0036] 2、 Real-time analysis of the speech content and interaction behavior of participants, when detecting theme attention deviation or new topic hotspots, automatically insert temporary path branches, so that the discussion always focuses on key issues, avoids decision lag or omission caused by deviating from the theme; by constructing a knowledge graph based on the attention labels and role state triples in multiple historical meetings, the role division strategy is continuously accumulated and optimized, which helps to form the best decision-making process and continuously improve;

[0037] 3、 Introducing a permission anomaly detection algorithm based on Mahalanobis distance and EWMA smoothing, monitoring the fine-grained permission vector after each role switching, and when an abnormal point occurs, rolling back or isolating the permission in time to prevent unauthorized access. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A flow chart of the intelligent paperless office management method based on the Internet of Things;

[0039] Figure 2 A structural schematic diagram of the intelligent paperless office management system based on the Internet of Things. DETAILED DESCRIPTION

[0040] To make the purpose, technical scheme and advantages of the present application clearer and more intelligible, the present application will be further described in detail below with reference to the specific embodiments and the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present application.

[0041] Embodiment 1: An intelligent paperless office management method based on the Internet of Things, as shown in Figure 1 , comprising the following steps:

[0042] S1: Obtain conference-related environmental data and pre-set conference agenda information before the meeting through the Internet of Things sensors deployed in the conference room;

[0043] S2: Based on the conference agenda information, construct an attention point and role linkage knowledge graph in combination with historical conference data, and plan the role switching order of each participant at different agenda nodes based on the knowledge graph before the meeting starts, to generate a preset planned role switching path;

[0044] S3: During the meeting, automatically or manually switch the roles of the participants according to the preset planning role switching path, assign the corresponding note templates and permissions to the roles, and monitor and adjust the temporary planning role switching path;

[0045] S4: Express the permission set after each role switching as a fine-grained permission vector, calculate the Mahalanobis distance based on the permission vector and the expected permission vector in the historical normal state, monitor the abnormal points and make adjustments;

[0046] S5: When the abnormal state is removed, use the permission recovery formula to recover the permissions;

[0047] S6: Synchronize the modifications made by each participant to the conference files in real time to the conference server, and save the modification results of each participant in the conference storage module in a versioned manner after the meeting ends; The versioned saving should include the modifier identifier, modification timestamp and change summary.

[0048] The conference-related environmental data includes noise level, temperature, humidity, light intensity, air quality in the conference room, and personnel distribution collected by seat sensors; The pre-meeting preset conference agenda information includes the content theme, stage time arrangement and required leading role type of each agenda node.

[0049] Install environmental monitoring sensor nodes, noise sensors on the top and four corners of the conference room, real-time collect sound pressure level data in the conference room; temperature and humidity sensors, measure and report indoor temperature and relative humidity; light sensors, record illumination level; air quality sensors, monitor PM2.5 concentration and CO2 concentration, arrange seat pressure sensors under each table and chair back to collect and calculate the distribution of participants, such as several seats being occupied; In the pre-meeting management platform, input the agenda node information of this meeting in a structured format, the main fields include: content theme: such as "project progress report", "demand discussion" and "risk assessment"; Stage time arrangement: the start time and duration of each agenda; Leading role type: specify the role type that should be assumed for each agenda node, such as "reporter", "reviewer" and "recorder", according to the above hardware deployment and preset agenda, complete the synchronous collection of conference environmental data and agenda information, provide basic data for subsequent role scheduling and knowledge graph construction.

[0050] The content of the document in which the participants in the historical meeting pay attention is labeled to record a triple, the triple contains a meeting file identifier, an attention label, and a role state, a focus and role linkage knowledge graph is constructed based on triples of a preset number of historical meetings, an optimal preset planned role switching path before the meeting is constructed according to the focus and role linkage knowledge graph, a temporary path branch is generated according to the focus and role linkage knowledge graph and the environment data related to the meeting in the process of the meeting, wherein the focus deviation is the degree of deviation of the attention degree of the current participant to the agenda theme from the historical benchmark; the focus and role linkage knowledge graph is stored in a graph database, nodes include file identifier nodes, attention label nodes, role state nodes, and switching path nodes, edges represent multi-dimensional associations between nodes; and a graph mining algorithm is used to identify high-frequency attention combinations and typical conversion modes; the temporary path branch is a local insertion of a role switching sub-process at a node of the preset planned role switching path, and only the local theme or hot spot is supplemented, deleted, or sequentially fine-tuned.

[0051] The generation process of the attention label is as follows: step 1: obtaining original data: ASR text, document highlights and annotations, search keywords, and meeting agenda text; step 2: candidate keyword / phrase extraction: using TextRank, RAKE, or a statistical-based TF-IDF to extract candidate keywords;

[0052] step 3: theme clustering or theme model: selecting LDA or BERT embedding + clustering to map candidate keywords to a theme set; step 4: label standardization: mapping themes to a predefined label set (or dynamically expanding the label library), and selecting the most matching label through vector similarity (cosine similarity); step 5: label scoring and confidence: calculating a confidence score for each label (based on frequency of occurrence, proportion of covered documents, and semantic similarity weighting);

[0053] Example: ASR transcription: "Regarding the demand change for product A, we need to adjust the return value of interface B" → candidate keywords {"demand change", "interface B", "return value"} → label mapping {"demand change"} (confidence 0.92).

[0054] Archive data of multiple completed meetings are extracted from the database, including: meeting document identification (such as document ID), document attention records during the meeting (participants' highlighting, annotating, searching behavior on electronic documents), and role status at the time (role type label). Each attention record is labeled to generate a triple of "meeting document identification-attention label-role status", such as: (Doc123, "demand change", "reviewer") or (Doc123, "progress report", "reporter"). All historical triples are aggregated to form a basic data set; a graph database (such as Neo4j) is used to store the above triples, and nodes and edges are constructed: document identification node: representing each meeting document; attention label node: representing topic keywords or hot topic labels; role status node: representing role type; switching path node: used for subsequent planning of role switching sequence fragments; multi-dimensional association edges are established between nodes: document identification-attention label: representing the hot topic marked in the document; attention label-role status: representing the participation of the historical role under the hot topic; role status-switching path: representing the ability to switch to the subsequent role from the role status; common switching sequences between common attention label sets and role status are identified using graph mining algorithms; before the meeting, according to the pre-meeting preset meeting agenda information of the current meeting, the content theme corresponding to the attention label of each agenda node is found; the most relevant historical role status sequence of the attention label is retrieved in the knowledge graph; according to the stage time arrangement, the role switching sequence of each participant is planned, and a complete "preset planned role switching path" is generated, for example: agenda 1 (demand discussion, 09:00-09:30): participant A (reporter) → participant B (reviewer) → participant C (recorder); agenda 2 (risk assessment, 09:30-10:00): participant B (reporter) → participant C (reviewer) → participant A (recorder); during the meeting, conference-related environmental data (noise, temperature, and personnel distribution) and participant interaction behavior (document highlighting and speaking record) are continuously collected; according to the historical distribution of attention labels in the graph database, the attention deviation of the current agenda theme is calculated: attention deviation = |real-time attention degree-historical benchmark attention degree|; when the attention deviation exceeds the preset threshold (such as deviation rate > 20%) or other preset trigger conditions (high environmental noise or new topics added during the meeting) are met, temporary path branches are dynamically generated, for example: in "agenda 1", if the "budget adjustment" topic heat suddenly increases, a temporary branch is generated at this node: … → participant C (temporary reviewer) → participant D (technical support) → return to the original preset process …; the generated temporary path branches are issued to each client, and the path node weight in the graph database is updated for subsequent optimization; the data source and calculation of real-time attention degree A_t: data source: keyword / entity appearance frequency and theme relevance score (speech_score) after speech-to-text (ASR);Interaction score (doc_score) generated by electronic document interaction behavior (highlight, comment, search and page dwell time); Interaction event score (ui_score) generated by client interaction (click, activate editing tools and share operation); Occupancy score (presence_score) generated by physical sensor (seat occupancy, seat pressure or presence facing microphone / camera); Fusion formula: A_t = w1 x speech_score + w2 x doc_score + w3 x ui_score + w4 x presence_score; Where w1 + w2 + w3 + w4 = 1;

[0055] The construction method of the historical reference attention degree is as follows: N times of meetings similar to the current agenda theme are selected from a preset number of historical meeting records, and the average attention score of each historical meeting on the corresponding agenda node is calculated The reference is a weighted average: Or an exponentially weighted moving average (EWMA) is used to give higher weight to recent meetings.

[0056] According to the attention point and role linkage knowledge graph, the optimal preset planning role switching path before the meeting is constructed, the attention deviation is detected during the meeting, and a temporary path branch is generated, which comprises: according to the attention point and role linkage knowledge graph, the optimal preset planning path is intelligently recommended based on the attention label distribution of similar historical meetings before the meeting; the deviation of the current attention label and the historical distribution is detected during the meeting, and a temporary path branch is automatically prompted or inserted; the attention flow record across sessions is formed after the meeting is completed; and a timeout rollback mechanism is set, which is to cancel all associated operations after role switching.

[0057] Based on the collected conference environment data and the preset conference agenda before the meeting, in the preset planning role switching path, the roles of the participants are automatically or manually switched according to the conference process, and each type of role is assigned a corresponding note template and permission, during the meeting, the speech content and interaction behavior of the participants are analyzed in real time, when the rule trigger of the temporary planning role switching path is detected, the temporary role switching instruction is dynamically generated and sent to the client, the note tool and the permission corresponding to the switched role; wherein the preset planning role switching path is the role switching sequence planned in advance for each participant at each agenda node before the meeting; the temporary planning role switching path is a complete role switching sequence that replaces the remaining part of the preset planning role switching path at one time based on the real-time detected attention deviation or other trigger factors during the meeting.

[0058] The intelligent recommendation of the optimal preset planning path specifically includes the following input data: current agenda topic tags, pre-meeting agenda metadata (duration, dominant role type), distribution of attention tags in historical meetings, high-frequency path weights in the knowledge graph, participant role capabilities and attendance; feature vectorization: constructing feature vectors for each historical meeting or path, such as: attention tag vectors (based on TF-IDF or pre-trained embeddings), path frequency, average response time, and role matching score; similarity measurement: calculating the similarity between the current agenda and historical records using cosine similarity or embedding-based distance metrics; and using a weighted scoring function: Score = α·sim_tags + β·norm_freq + γ·role_match―δ·

[0059] `est_time_conflict`; where `Score` is the score value; `sim_tags` is the similarity between the current agenda and the historical records calculated using cosine similarity; `norm_freq` is the normalized path occurrence frequency; `role_match` is the matching degree between the role assignment in the path and the actual participants' abilities / identities; `est_time_conflict` is the estimated time conflict or timeout penalty; `α`, `β`, `γ`, and `δ` are adjustment weights; candidate paths are sorted in descending order of `Score`, and the top K paths are output as intelligent recommendation paths, with the first path being automatically adopted or adopted after confirmation by the administrator / host;

[0060] The norm_freq is statistically derived from the historical path database and normalized after time decay and Laplace smoothing. The role_match is the matching score obtained by matching the demand vector (skills, job level, and availability) of each role in the path with the current participant profile and then weighting and aggregating them according to node importance. The est_time_conflict is obtained by estimating the probability or expected timeout of the total path time exceeding the agenda allocation time based on the historical time distribution of each node in the path and normalizing it. Incremental correction of the time estimate based on real-time attention is supported during the real-time meeting.

[0061] After the meeting, the actual attention tags, role switching sequence and triggered temporary path branch information in each stage of the meeting are written in the form of a triple of "attention tag-role state-path node" into the cross-session log; these cross-session attention flow records are used for subsequent historical meeting retrieval, model training and path optimization; during the meeting, if no expected operation corresponding to the role switching is detected within the preset timeout time (e.g. 5 minutes) after the role switching is performed, such as no record or expected interaction tool is triggered, a timeout rollback mechanism is triggered, which will undo all associated operations after the role switching, including: the role state is restored to the state before the switching, the note template and the permission are restored to the original preset configuration, and the client interface is synchronized to ensure consistent expected process consistency; combined with the "meeting related environment data" and "pre-meeting meeting agenda information", when each agenda node is reached, the role state of the corresponding participant is switched automatically or through manual confirmation by the administrator according to the preset planned role switching path, and at the time of switching, the role switching instruction is issued to the client, and the note template and fine-grained permission vector corresponding to the role are loaded, for example: when the agenda transitions from "demand report" to "technical review", participant B is automatically switched from "recorder" to "reviewer", and the reviewer's special marking and commenting permissions are issued; during the meeting, the speech content, document highlighting and annotation interaction behavior of the participants are continuously collected and mapped as attention tags, and at the same time, the attention degree of the current agenda theme is evaluated in real time combined with environmental sensor data (such as seat occupancy or noise change).

[0062] The method for monitoring abnormal points and making adjustments is as follows,

[0063] The permission set after each role switching is represented as a vector P i =[p i,1 ,p i,2 ……,p i,m ] T , where p i,j ∈{0,1} represents whether the jth fine-grained permission is possessed after the ith role switching; m is the total number of all fine-grained permissions defined by default; the Mahalanobis distance square is obtained using the Mahalanobis distance formula, and the smoothed statistic is obtained by EWMA smoothing of the Mahalanobis distance; when the Mahalanobis distance square is greater than or equal to the preset threshold, this role switching is regarded as an abnormal point, and transaction rollback or permission isolation is performed; when the abnormal state is removed, the permission is restored using the permission recovery formula, where the Mahalanobis distance formula is:

[0064]

[0065] In the formula, is the Mahalanobis distance square; P iis the fine-grained permission set vector held by the current user after the i-th role switching; μ i is the expected value of the permission vector in the historical normal switching process under the i-th role switching stage; ∑ i is the covariance matrix of the fine-grained permission set vector under the i-th role switching stage;

[0066] denotes the inverse of the covariance matrix; () T is the transpose.

[0067] m fine-grained permissions are predefined, such as document reading, comment highlighting, initiating sharing, and voting, and the total number is m; after each role switching is completed, the server constructs a binary vector according to the permission set that should be issued for the role, where p i,j = 1 indicates that the user has the j-th permission after the i-th role switching, and p i,j = 0 indicates that the user does not have the j-th permission; in order to suppress the influence of single fluctuation, an exponentially weighted moving average smoothing statistic S i is introduced: where S0 can be initialized as the historical average Mahalanobis distance, where transaction rollback means to cancel the role switching action this time; permission isolation means to narrow the user's permissions to only the minimum permission set to prevent the spread of security risks.

[0068]

[0069] In the formula, is the permission vector that should be issued to the user after the abnormality is detected under the i-th role switching stage; μ i is the expected value of the permission vector in the historical normal switching process under the i-th role switching stage; α ′ is the recovery coefficient based on EWMA smoothing; β is the temporary path influence weight, β ∈ [0, 1], and the larger the value, the more the history of the temporary branch is valued, where α ′ + β ≤ 1; ΔB i is the average deviation vector of the expected temporary branch of the historical preset number, reflecting the contribution direction and strength of the temporary path to the permission allocation; P min is the minimum permission set, which includes the minimum permission that guarantees the basic participation ability of the meeting; α ′ is the recovery coefficient; z is the preset threshold; S i is the smoothing statistic.

[0070] The average deviation vector of the expected temporary branch of the historical preset number is obtained through the temporary vector calculation formula, where the temporary vector calculation formula is:

[0071]

[0072] wherein, ΔB i is the temporary branch of the historical preset number of times and the expected average deviation vector; K is the historical preset number of times; is the actual authority vector corresponding to the i-th role switching stage of the k-th temporary switching branch; μ i is the expected value of the authority vector in the historical normal switching process under the i-th role switching stage.

[0073] The difference between the actual authority vector actually issued due to the temporary path branch and the expected authority vector is averaged after being statistically counted for K times history; the recovery coefficient is adaptively generated according to the following formula in combination with the smoothing statistic and the preset threshold When S i is large, α ′ is close to 0, and more dependent on the temporary deviation and the minimum authority; if S i is small, α ′ is close to 1, and more dependent on the historical expectation; Round() represents the final binary authority obtained by rounding each component in the vector.

[0074] Before the meeting, the participants are arranged in time periods according to the meeting agenda and the participants' schedule, and the files required for the meeting are pre-issued and cached to each participant terminal; after the meeting, the modified content of each participant is summarized and saved in the conference storage module.

[0075] In this embodiment, the Internet of Things device is a tablet or a mobile phone, and the power, memory, offline or online, and specific information content of the user are monitored in real time, and the idle status and detailed information of the borrowed tablet or mobile phone are displayed.

[0076] Embodiment 3: Based on the embodiments 1 and 2, an intelligent paperless office management system based on the Internet of Things, as shown in Figure 2 , further comprises:

[0077] The data acquisition module is configured to acquire conference-related environmental data and pre-set conference agenda information through the Internet of Things sensors deployed in the conference room.

[0078] The path construction module is configured to construct a concern point and role linkage knowledge graph based on the conference agenda information and historical conference data, and plan the role switching order of each participant at different agenda nodes based on the knowledge graph before the conference starts, and generate a pre-set planned role switching path.

[0079] The monitoring module switching module is configured to automatically or manually switch the participant roles according to the pre-set planned role switching path during the conference, and assign the corresponding note templates and authorities to the roles, and monitor and adjust the temporary planned role switching path.

[0080] Anomaly monitoring module, for representing the permission set after each role switch as a fine-grained permission vector, and calculating Mahalanobis distance based on the permission vector and the expected permission vector in the historical normal state, monitoring the anomaly point and adjusting;

[0081] Permission recovery module, for recovering the permission using the permission recovery formula when the abnormal state is removed;

[0082] Data storage module, for synchronizing the modifications made by each participant to the conference file in real time to the conference server during the conference, and saving the modification results of each participant in a versioned manner in the conference storage module after the conference ends; the versioned saving should include the modifier identification, modification timestamp and change summary.

[0083] The above embodiments are provided for those skilled in the art to implement or use the present application, those skilled in the art can make various modifications or changes to the above embodiments without departing from the inventive concept of the present application, therefore the protection scope of the present application is not limited by the above embodiments, but should be the maximum scope meeting the innovative features mentioned in the claims.

Claims

1. A smart paperless office management method based on the Internet of Things, characterized in that, Includes the following steps: S1: Obtain environmental data related to the meeting and pre-set meeting agenda information through IoT sensors deployed in the meeting room; S2: Based on the meeting agenda information, construct a knowledge graph of the linkage between focus and role by combining historical meeting data, and plan the role switching order of each participant at different agenda nodes based on the knowledge graph before the meeting starts, and generate a preset planned role switching path; S3: During the meeting, the participant roles are automatically or manually switched according to the preset role switching path, and corresponding note templates and permissions are assigned to the roles. The system is also monitored and adjusted to a temporary role switching path. S4: Represent the set of permissions after each role switch as a fine-grained permission vector, calculate the Mahalanobis distance between this permission vector and the expected permission vector under the historical normal state, monitor anomalies and make adjustments; S5: After the abnormal state is resolved, use the permission recovery formula to restore permissions; S6: Synchronize the modifications made to the meeting documents by each participant during the meeting to the meeting server in real time, and save the modifications made by each participant in a versioned manner in the meeting storage module after the meeting ends; The versioned storage should include the modifier's identifier, modification timestamp, and change summary.

2. The intelligent paperless office management method based on the Internet of Things as described in claim 1, characterized in that, The method involves acquiring meeting-related environmental data and pre-set meeting agenda information through IoT sensors deployed in the conference room. The meeting-related environmental data includes noise levels, temperature, humidity, light intensity, air quality, and personnel distribution data collected by seat sensors in the conference room. The pre-set meeting agenda information includes the content theme, time schedule, and required leading role type for each agenda node.

3. The intelligent paperless office management method based on the Internet of Things as described in claim 2, characterized in that, The process involves constructing a knowledge graph linking attention points and roles based on the meeting agenda information and historical meeting data. Before the meeting begins, the system plans the role switching order for each participant at different agenda nodes based on this knowledge graph, generating a preset role switching path. This includes: tagging the content of documents that participants paid attention to in historical meetings and recording them as triples. Each triple contains a meeting document identifier, attention tag, and role status. Based on a preset number of historical meeting triples, the system constructs a knowledge graph linking attention points and roles. Based on this knowledge graph, the system constructs the optimal preset role switching path before the meeting. During the meeting, based on the knowledge graph and meeting-related environmental data, the system detects attention deviations and generates temporary path branches. The attention deviation is the degree to which the current participant's attention to the agenda topic deviates from the historical baseline. The knowledge graph is stored in a graph database, with nodes including document identifier nodes, attention tag nodes, role status nodes, and switching path nodes. Edges represent multi-dimensional relationships between nodes. Graph mining algorithms are used to identify high-frequency attention combinations and typical switching patterns; the temporary path branch is a sub-process of role switching that is partially inserted at a certain node of the preset planned role switching path, which only adds, deletes or fine-tunes the order of roles for the local topic or hotspot.

4. The intelligent paperless office management method based on the Internet of Things as described in claim 1, characterized in that, The process of constructing an optimal pre-planned role switching path based on the knowledge graph of attention points and role linkage before the meeting, and detecting attention deviations and generating temporary path branches during the meeting based on the knowledge graph of attention points and role linkage, includes: intelligently recommending the optimal pre-planned path based on the distribution of attention tags in similar historical meetings before the meeting; detecting deviations between the current attention tags and historical distributions during the meeting and automatically prompting or inserting temporary path branches; forming a cross-session attention flow record after the meeting; and setting a timeout rollback mechanism to undo all associated operations after role switching.

5. A smart paperless office management method based on the Internet of Things as described in claim 4, characterized in that, The process of automatically or manually switching participant roles according to the preset role switching path during the meeting, assigning corresponding note templates and permissions to each role, and monitoring and adjusting to a temporary role switching path includes: based on the collected meeting environment data and the pre-set meeting agenda, automatically or manually switching participant roles according to the meeting flow within the preset role switching path, assigning corresponding note templates and permissions to each role, analyzing participant speech content and interaction behavior in real time during the meeting, and dynamically generating a temporary role switching instruction and sending it to the client when a rule matching the temporary role switching path is detected, switching the note-taking tools and permissions corresponding to the role; wherein the preset role switching path is the role switching sequence pre-planned for each participant at each agenda node before the meeting begins; the temporary role switching path is a complete role switching sequence that replaces the remaining part of the preset role switching path in one go during the meeting, based on real-time detected attention deviations or other triggering factors.

6. The intelligent paperless office management method based on the Internet of Things as described in claim 1, characterized in that, The step of representing the permission set after each role switch as a fine-grained permission vector, calculating the Mahalanobis distance between this permission vector and the expected permission vector under historical normal conditions, monitoring anomalies and making adjustments includes: representing the permission set after each role switch as a vector P. i =[p i,1 ,p i,2 ...,p i,m ] T , where p i,j ∈{0,1} indicates whether the j-th fine-grained permission is possessed after the i-th role switch; m is the total number of all predefined fine-grained permissions; the squared Mahalanobis distance is obtained using the Mahalanobis distance formula, and EWMA smoothing is applied to the Mahalanobis distance to obtain a smoothing statistic. When the squared Mahalanobis distance is greater than or equal to a preset threshold, this role switch is considered an anomaly, and transaction rollback or permission isolation is performed. After the anomaly is resolved, the permission is restored using the permission recovery formula, where the Mahalanobis distance formula is: In the formula, P is the squared Mahalanobis distance; i Let μ be the vector of fine-grained permissions held by the current user after the i-th role switch; i Let ∑ be the expected value of the permission vector during the historical normal switching process at the stage of the i-th role switching; i Let be the covariance matrix of the fine-grained permission set vector under the i-th role switching stage; Represents the inverse of the covariance matrix; () T This is a transpose.

7. The intelligent paperless office management method based on the Internet of Things as described in claim 1, characterized in that, The step of restoring permissions using the permission restoration formula after the abnormal state is resolved includes: where the permission restoration formula is: In the formula, During the i-th role switch phase, after anomaly detection, the permission vector to be issued to the user is restored; μ i Let α be the expected value of the permission vector during the historical normal switching process at the stage of the i-th role switch; ′ α represents the recovery coefficient based on EWMA smoothing; β is the weight of the transient path influence, β∈[0,1], a larger value indicates greater emphasis on the historical influence of transient branches, where α ′ +β≤1;ΔB i The average deviation vector between the historical preset number of temporary branches and the expected value reflects the direction and strength of the temporary path's contribution to permission allocation; P min This is the minimum set of permissions, containing the minimum permissions required to guarantee basic participation in the meeting; α ′ is the recovery coefficient; z is the preset threshold; S i For smoothing statistics.

8. A smart paperless office management method based on the Internet of Things as described in claim 7, characterized in that, The average deviation vector between the historical preset number of temporary branches and the expected value includes: obtaining the average deviation vector between the historical preset number of temporary branches and the expected value through a temporary vector calculation formula, wherein the temporary vector calculation formula is: In the formula, ΔB i K represents the average deviation vector between the temporary branches and the expected values ​​after a historical preset number of iterations; K is the historical preset number of iterations. μ represents the actual permission vector corresponding to the k-th temporary switch branch during the i-th role switch phase; i Let be the expected value of the permission vector during the historical normal switching process at the stage of the i-th role switching.

9. A smart paperless office management method based on the Internet of Things as described in claim 1, characterized in that, The modification made by each participant to the meeting documents during the meeting is synchronized to the meeting server in real time, and the modification results of each participant are saved in a versioned manner in the meeting storage module after the meeting ends; The versioned storage should include the modifier's identifier, modification timestamp, and change summary, including: Before the meeting, participants are scheduled in time slots according to the meeting agenda and their schedules, and the necessary documents are pre-distributed and cached on each participant's terminal. After the meeting, the modifications made by each participant are summarized and saved in the meeting storage module.

10. An IoT-based intelligent paperless office management system, used to implement the IoT-based intelligent paperless office management method according to any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire environmental data related to the meeting and pre-set meeting agenda information through IoT sensors deployed in the meeting room; The path construction module is used to construct a knowledge graph of the linkage between the focus and the role based on the meeting agenda information and historical meeting data, and to plan the role switching order of each participant at different agenda nodes based on the knowledge graph before the meeting starts, and generate a preset planned role switching path. The monitoring module switching module is used to automatically or manually switch the participant roles according to the preset planned role switching path during the meeting, assign corresponding note templates and permissions to the roles, and monitor and adjust to a temporary planned role switching path. The anomaly detection module is used to represent the set of permissions after each role switch as a fine-grained permission vector, calculate the Mahalanobis distance based on this permission vector and the expected permission vector under the historical normal state, detect anomalies and make adjustments accordingly. The permission recovery module is used to restore permissions using permission recovery formulas after an abnormal state is resolved. The data storage module is used to synchronize the modifications made by each participant to the meeting documents during the meeting to the meeting server in real time, and to save the modifications made by each participant in a versioned manner in the meeting storage module after the meeting ends; The versioned storage should include the modifier's identifier, modification timestamp, and change summary.

Images