A node claim, relationship merging and intelligent service method and system based on family gene chain

Abstract

The application discloses a kind of node claim based on family gene chain, relationship merging and intelligent service method and system.The method includes: with the node of user oneself as center, through recursive expansion and the merging of shortest path based on the degree of repetition node generated in expansion and rejudgment, form dynamic scalable family gene chain;Through presetting relative node, the matching confidence calculation based on multi-source evidence fusion and the confirmation of relationship both sides, the new registered user node is merged with its corresponding preset relative node;Within the range of limited relative degree, according to application scene, relevant node is screened and its desensitization after graph label is called, and family scene intelligent service is executed.The application can realize the safe claim and cross-network automatic merging of unregistered relatives, provide intelligent question and answer, push, event recommendation, photo album classification and title query services for real family relationship under the premise of protecting privacy, significantly improve the accuracy and usability of family scene intelligent service.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence application technology, and in particular to a method and system for node adoption, relationship merging and intelligent services based on family gene chains. Background Technology

[0002] With the development of digital social networking and family services, various relationship management systems are emerging. However, existing contact systems only record "who is connected to whom," failing to express kinship semantics such as paternal / maternal lineage, blood / marriage, seniority, and generational relationships; traditional genealogy systems focus on static display and lack the ability to build dynamically expandable relationship networks centered on individuals; social systems rely on mutual friend requests to establish connections, unable to pre-establish links before relatives register and automatically claim and merge them after registration; general artificial intelligence models lack an understanding of users' real family environments, making it difficult to provide intelligent results that fit kinship relationships in scenarios such as holiday reminders, gift recommendations, and family event services. Summary of the Invention

[0003] In view of the above-mentioned shortcomings in the current field of artificial intelligence application technology, the present invention provides a method and system for node adoption, relationship merging and intelligent services based on family gene chain. It can realize a secure closed loop of building the chain first and then adopting the node, avoid misclassification of duplicate individuals, support automatic cross-network merging, and provide AI-based Q&A, intelligent push, event recommendation, photo album classification and title query services for real family relationships while protecting privacy. This significantly improves the accuracy and usability of intelligent services in family scenarios.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions: A method for node adoption, relationship merging, and intelligent services based on family gene chains includes: Centered on the user's own node, a dynamic and scalable family gene chain is formed by recursively expanding and performing shortest path-based merging and degree re-evaluation on duplicate nodes generated during the expansion. By pre-setting kinship nodes, calculating matching confidence based on multi-source evidence fusion, and confirming the relationship between the two parties, the newly registered user node is merged with its corresponding pre-set kinship node; within the limited kinship degree range, relevant nodes are selected according to the application scenario and their desensitized graph tags are called to execute intelligent services for family scenarios.

[0005] According to one aspect of the present invention, the construction of the family gene chain specifically includes: taking the user's own node as the zero-degree node, recursively expanding the relationship edges around five types of relationships: father, mother, spouse, sibling, and children, to form a candidate node set; performing duplicate person identification and node merging on the candidate node set, and calculating the shortest relationship path length of each merged node relative to the user's own node; only when the shortest path length is equal to the target degree is the node classified into that degree layer, otherwise its original lower degree classification is maintained.

[0006] According to one aspect of the present invention, the preset relative node includes: when the target relative has not yet registered, a preset relative node is created by a registered user, the node occupies a certain position in the relationship graph and records the relationship type, basic identity information and pending confirmation status; when the target relative subsequently registers through a mobile phone number to form a new registered user node, the system retrieves the matching preset relative node and generates a matching confidence score.

[0007] According to one aspect of the present invention, the matching confidence calculation based on multi-source evidence fusion adopts the following formula: C=100×(0.45I+0.20B+0.35T)−P Wherein, I is the score for contact information and registration source evidence, B is the basic information consistency score, T is the relationship topology consistency score, and P is the conflict penalty score; the contact information and registration source evidence includes consistent mobile phone number, consistent de-identified mobile phone number characteristics, or invitation link binding; the basic information consistency includes name, date of birth or age group, and gender; the relationship topology consistency includes relationship role consistency, shortest path degree consistency, kinship topology legality, and relationship closure gain; the conflict penalty score is taken from 0 to 100 based on user rejection, real name conflict, conflict between direct blood relatives and spouses, gender and relationship conflict, age and generational conflict, and inability to calculate the title path, etc., where merging is prohibited when P=100.

[0008] According to one aspect of the present invention, the conditions for confirmation and node merging between the two parties in the relationship are as follows: when the matching confidence C ≥ 85, the system initiates confirmation between the identified relative and the relative party; the system merges the newly registered user node with the original preset relative node only when the identified relative confirms, the relative party confirms, there is no hard conflict, and the shortest path gradation after merging is correct; when C < 70, no merging is prompted.

[0009] According to one aspect of the present invention, it further includes a network merging process after two registered users confirm their relationship: after two registered users complete the relationship confirmation, the system uses the two newly registered user nodes as network merging anchors, generates candidate merging relationships around the existing preset family nodes of both parties, and calculates the confidence of each candidate group one by one according to the relationship mapping rules and the matching confidence model; for candidate nodes that reach the merging threshold and have no conflict, confirmation or merging is initiated; for low-confidence candidate nodes, only the candidate association status is retained.

[0010] According to one aspect of the invention, it further includes binding a personal knowledge graph to each individual node in the family gene chain: the personal knowledge graph includes multiple pieces of information such as basic attributes, interests and preferences, behavioral habits, health taboos, consumption style, family roles, relational memories, emotional triggers, and behavioral feedback; recording the data source, timestamp, version number, and credibility parameter for each graph field; assigning different credibility to data from different sources, retaining historical versions, and performing weight decay processing on new and old preferences; when information from different sources conflicts, prioritizing the version with higher credibility and more recent time.

[0011] According to one aspect of the present invention, the intelligent service for family scenarios further includes standard title query: obtaining the shortest relationship path from the user node to the target node, reading the relationship type, generational direction, paternal or maternal lineage, blood or in-laws and seniority attributes of each relationship edge on the path, and outputting a unified standard title according to a preset title mapping rule; the user cannot directly edit the title result, but indirectly corrects it by modifying the relationship topology.

[0012] According to one aspect of the present invention, the smart home service further includes at least one of the following: AI-powered emotional Q&A: When users initiate questions about family relationships, gift-giving, communication, holiday greetings, or expressions of affection, the system generates answers, communication suggestions, or action suggestions tailored to family relationships based on the user's relationship path with the target relatives, standard titles, and desensitized map tags. Intelligent push notification: When the system actively identifies family events such as birthdays, holidays, anniversaries, weddings, birthday celebrations, family gatherings, elder care, or kinship interactions, it determines the relationship path, standard titles, degree of kinship, and family roles between the key figures of the event and the push recipients based on the family gene chain. It then calls the desensitized map tags of the key figures of the event, the push recipients, and related relatives to generate differentiated push content with emotional expression. Finally, it matches at least one product, service, virtual item, blessing, small red envelope, local life service, or online mall entrance and pushes it to the corresponding user. Family album categorization: Based on the event tags of the photos and the relationship path of the identified person nodes in the family gene chain, the photos are automatically categorized into the corresponding relationship branch albums; Standard title query: Obtain the shortest relationship path from the user node to the target node, read the relationship type, generation direction, paternal or maternal lineage, blood or in-laws, and seniority attributes of each relationship edge on the path, and output a unified standard title according to the preset title mapping rules.

[0013] A family gene chain node claiming and intelligent service system includes: a family chain construction module, which takes the user's own node as the center, and forms a dynamically scalable family gene chain by recursively expanding and performing shortest path-based merging and degree re-judgment on duplicate nodes generated in the expansion. The node and confidence module merges the newly registered user node with its corresponding preset family node through preset family node, matching confidence calculation based on multi-source evidence fusion, and confirmation from both parties to the relationship; The merging and confirmation module, within a limited range of kinship degrees, filters relevant nodes according to the application scenario and calls their desensitized graph tags to execute intelligent services for the family scenario; The desensitization module is used to execute intelligent family scene services by calling desensitized map tags within a limited range of relatives' degrees, based on the scenario. The storage module is used to store the multidimensional semantic graph information of each character node, as well as its source, version, and credibility parameters.

[0014] The advantages of this invention are: it enables the construction of a dynamic family relationship network centered on the individual, recursively expandable, and with accurate degrees; it achieves safe and controllable kinship recognition and automatic cross-network merging through preset nodes and a multi-evidence confidence model; and it calls upon the individual's knowledge graph based on de-identified tags within a four-degree range. Thus, while protecting user privacy, it provides a real, continuous, and verifiable kinship background for family-oriented services such as intelligent question answering, personalized push notifications, family event recommendations, photo album categorization, and standard title queries, significantly improving the accuracy, contextual relevance, and usability of the services. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a method for node adoption, relationship merging, and intelligent services based on a family gene chain, as described in this invention. Figure 2 This is a schematic diagram illustrating the principle of kinship degree formation as described in this invention; Figure 3This is a schematic diagram illustrating the mother node expansion, deduplication, and re-calibration described in this invention; Figure 4 This is a flowchart of the preset node claiming and relationship network merging process described in this invention; Figure 5 This is a schematic diagram of the standard title query path described in this invention; Figure 6 This is a schematic diagram illustrating the joint use of multiple character atlases as described in this invention; Figure 7 This is a flowchart illustrating the node recognition and intelligent service system for a family gene chain as described in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 like Figure 1 As shown, a method for node adoption, relationship merging, and intelligent services based on family gene chains includes the following steps: Step S1: Centered on the user's own node, expand recursively and perform shortest path-based merging and degree re-evaluation on duplicate nodes generated during the expansion to form a dynamically scalable family gene chain. S101: Create your own core node and assign it a unique account identifier, node identifier, and initial relationship attributes.

[0019] S102: The user adds first-degree relatives nodes through the relationship topology graph interaction interface; the first-degree relatives nodes include father nodes, mother nodes, spouse nodes, sibling nodes and child nodes; first-degree relatives are determined according to the shortest path length to the user's node being 1.

[0020] S103: For any person node in the generated degree layer, continue to use its father, mother, spouse, siblings and children as the entry points for expanding the relationship, forming the next round of candidate node set.

[0021] S104: Duplicate Person Recognition and Node Merging. After the candidate node set is formed, the system performs duplicate person recognition and node merging, which includes the following sub-steps: S1041: Generate candidate duplicate node pairs. During the recursive expansion, preset node claiming, or network merging process after two users confirm their relationship, the system discovers nodes that may point to the same real person and generates candidate duplicate node pairs. Candidate node pairs include: duplicate nodes generated in the same path expansion, nodes from different paths pointing to the same relative role, nodes of the same relative added by different users, and candidate corresponding nodes derived from the network merging anchor point.

[0022] S1042: Determine the candidate triggering conditions. When a candidate node pair meets any of the following conditions, the system enters the duplicate person recognition process: The phone numbers on both nodes are the same; The two nodes are associated through the same invitation link or the same preset node source; The two nodes have the same name, and the same date of birth or age range; The two nodes have the same relationship path; The two nodes point to a common relative; The two nodes are derived from the same confirmed network anchor point; Two nodes form a closable path in the relational topology.

[0023] If two nodes have the same name but different phone numbers, birthdays, relationship paths, and common relatives, they will not enter the automatic merging process.

[0024] S1043: Calculate the matching confidence score. The system calculates the matching confidence score based on contact information and registration source evidence, consistency of basic information, consistency of relationship topology, and conflict penalty term. The matching confidence score formula is: C = 100 × (0.45I + 0.20B + 0.35T) - P in: C represents the match confidence level, with a value ranging from 0 to 100; I represents the score for evidence of contact information and registration source (consistent phone number, binding of invitation link, etc.); The rules for determining the values ​​of contact information and proof of registration source I are as follows: The phone numbers are exactly the same: I=1.00; Register via the invitation link linked to this preset node: I=0.90; If the last four digits of the phone number and the name match: I=0.65; Name only, no phone number or proof of origin: I=0.00; If the phone number is clearly inconsistent and there is no other strong evidence: proceed to conflict penalty.

[0025] B is the basic information consistency score (name, date of birth, gender); The rules for determining the value of Basic Information Consistency B: B = 0.50×B1 + 0.35×B2 + 0.15×B3 B1 (Name Consistency): Complete Consistency = 1.00, Partial Consistency = 0.60, Missing Information = 0.30, Explicit Conflict = 0; B2 (Date of Birth or Age Group Consistency): Complete Consistency = 1.00, Partial Consistency = 0.60, Missing Information = 0.30, Explicit Conflict = 0; B3 (Gender Consistency): Consistency = 1.00, Missing Information = 0.30, Significant Conflict with Relationship Role = 0 (and enters conflict penalty).

[0026] T represents the relation topology consistency score (relation role consistency, shortest path degree consistency, kinship topology legitimacy, relation closure gain); T = 0.40×T1 + 0.30×T2 + 0.20×T3 + 0.10×T4 T1 (Relationship Role Consistency): Default value: Complete match between the relationship role of the node and the registered user = 1.00, Partial match = 0.60, No match = 0; T2 (Shortest Path Consistency): After temporary merging, the shortest path length is correct = 1.00, incorrect = 0; T3 (Family Topological Legality): Merging does not violate kinship logic (e.g., no parent-child reversal, direct line and spouse conflict, etc.) = 1.00, conflict exists = 0; T4 (Relation Closure Gain): If a reasonable new relation closure can be formed after merging, =1.00; otherwise =0.50.

[0027] P represents the conflict penalty score (user rejection, real-name conflict, conflict between direct blood relatives and spouses, age and generational conflict, etc.; merging is prohibited when P=100).

[0028] S1044: Conflict veto and mutual confirmation. The system outputs the following results based on the confidence level C and the conflict status: If C ≥ 85, and both parties confirm there is no hard conflict: the system merges candidate nodes; C ≥ 85, but mutual confirmation is not completed: The system generates a strong candidate relationship and initiates relationship confirmation to the relevant users; 70 ≤ C < 85: The system generates weak candidate relationships, only prompting or retaining the background candidate status; C < 70: The system does not prompt for merging; A hard conflict exists (P=100): the system prohibits merging.

[0029] The final merge must simultaneously meet the following conditions: the confidence level reaches the threshold, the identified relative confirms the identity, the related party confirms the identity, there is no hard conflict, and the shortest path metric after merging is correct.

[0030] S1045: Processing after node merging. When the system confirms that two or more candidate nodes belong to the same real person, it performs node merging and relationship refresh: merging the candidate nodes into the same person's main node. Preserve the relationship edges between each source node; Merge the basic information fields and record the source of each field; Update the node status to confirmed node or shared preset node; Recalculate the shortest path, degree of kinship, and standard title of the person node relative to the relevant users; Refresh the visibility range of the four-degree relationship and the access permissions for calling smart services.

[0031] S105: Only when the shortest path length of a person node relative to its own node is equal to the target degree, will the person node be classified into the target degree level; if the shortest path length is shorter, it will retain its original lower degree classification.

[0032] S106: During the expansion process, record one or more of the following for each relation edge: relation type, generational direction, paternal or maternal lineage, blood or in-law relationship, seniority, source node, and confirmation status.

[0033] S107: When a user adds a relative's location incorrectly, the relationship topology is corrected by deleting the incorrect node and re-adding it in the correct location, rather than by directly modifying the title result.

[0034] S108: The system can continue to recursively expand in the direction of ancestors and descendants to form a deeper family relationship network; however, the visible relationship range open to ordinary users is limited to within four degrees.

[0035] Through the above steps, with the user's own node as the center, a dynamic and scalable family gene chain is formed by recursively expanding and performing shortest path-based merging and degree re-evaluation on duplicate nodes generated during the expansion.

[0036] Step S2: Merge the newly registered user node with its corresponding preset family node by using preset family node, calculating the matching confidence based on multi-source evidence fusion, and confirming the relationship between the two parties; S201: When the target relative has not yet registered, a preset relative node is created by a registered user; the preset relative node occupies a definite position in the relationship graph, but the real identity binding has not yet been completed.

[0037] S202: The system saves the relationship location, basic identity information, creation source, and pending confirmation status of the preset relative node for subsequent matching and claiming.

[0038] S203: When a pre-defined relative completes registration and forms a new registered user node, the system searches for a pre-defined relative node that highly matches the newly registered user.

[0039] S204: If a highly matching node exists, display a candidate relationship suggestion to the newly registered user and ask whether the candidate relationship matches their real family relationship.

[0040] S205: The system calculates the matching confidence level based on a combination of primary and secondary evidence. Primary evidence includes at least one of the following: phone number matching, invitation link binding, registration source binding, or preset node source association; real-name information is an optional verification item and is not a necessary condition for ordinary user registration, chain building, or node claiming. If the user voluntarily provides real-name information and the real name is consistent, it can be used as a basis for enhanced verification; if the real-name information is clearly conflicting, it is considered a hard conflict and automatic merging is prohibited. Secondary evidence includes at least one of the following: name and birthday matching, relationship path consistency, and cross-identification by multiple people.

[0041] S206: Generate matching confidence results based on the primary and secondary evidence, and identify whether the candidate nodes meet the conditions for identifying the same relative.

[0042] S207: Provided that the candidate relationship is met and the matching confidence level reaches the threshold, both users on both sides of the relationship shall confirm the relationship; after both parties confirm the relationship, the system shall merge the newly registered account node with the original preset family node.

[0043] S208: After automatic merging, the system generates a closable candidate relationship based on the confirmed shared relative nodes, and performs relationship mapping, matching confidence calculation and topological conflict verification on each candidate node; the candidate node merging is only performed when the candidate node reaches the merging threshold and completes the necessary confirmation, so as to avoid erroneously merging non-direct relatives such as step-relatives and remarried relatives.

[0044] S209: If a newly registered user refuses to claim the relationship, the two sides of the relationship have not completed confirmation, there is conflicting evidence, the path logic is inconsistent, or the confidence result is insufficient, then the node will remain in a separate state, the relationship network will not be merged, and only the candidate state will be retained.

[0045] Through the above steps, by pre-setting family nodes, calculating the matching confidence based on multi-source evidence fusion, and confirming the relationship between the two parties, the newly registered user node is merged with its corresponding pre-set family node.

[0046] Step S3: Within the defined range of kinship degrees, select relevant nodes according to the application scenario and call their desensitized graph tags to execute intelligent services for the family scenario.

[0047] S301: Bind a corresponding personal knowledge graph to each character node in the family gene chain. The personal knowledge graph at least includes five types of label sets: physiological characteristics, personality and behavior, life details, family relationships, and emotional characteristics. The graph data sources include user manual input, family member supplementation, chat texts, speech-to-text, browsing records, purchase records, likes and collections, festival behaviors, gift feedback, and AI-inferred labels, etc. Record the data source, timestamp, version number and credibility parameter for each graph field, assign different credibility levels to data from different sources, retain historical versions, and perform weight attenuation processing on old and new preferences.

[0048] The weight attenuation processing adopts the following formula: wherein, is the current weight, is the initial weight, is the time attenuation coefficient, is the time interval from the data generation time to the current time, is the source credibility coefficient. The credibility of the source of information actively filled in by the person himself / herself is higher than that of AI-inferred information; the weight of recent high-frequency behaviors is higher than that of early low-frequency behaviors; the weight of information consistently confirmed by multiple relatives is higher than that of information inferred from a single source.

[0049] S302: Standard address calculation. The system takes the shortest relationship path from the self node to the target node as the basis for address calculation. It specifically includes the following sub-steps: S3021: Acquire the shortest relationship path. The system starts from the self node and acquires the shortest relationship path from the self to the target character node. When there are multiple reachable paths for the same target character, the shortest path is preferentially adopted; if there are multiple shortest paths of equal length, the confirmed relationship path, blood relationship path, and direct relationship path are preferentially adopted; if there are still multiple address candidates, secondary determination is performed according to gender, seniority, marital relationship, and confirmation status.

[0050] S3022: Generate a relationship path code. The system converts the relationships of father, mother, spouse, sibling, and child in the path into single-character Chinese codes respectively: "father, mother, spouse, sibling, child". For example: The path "Self → Mother → Sibling → Child → Spouse" is coded as: Mother-Sibling-Child-Spouse; The path "Self → Father → Sibling → Child" is coded as: Father-Sibling-Child; The path "Self → Spouse → Father" is coded as: Spouse-Father; The path "Self → Child → Child" is coded as: Child-Child.

[0051] S3023: Read auxiliary attributes. The system reads the following auxiliary attributes of the target node and key nodes in the path: gender (male / female / unknown), age (older / younger / unknown), lineage (paternal / maternal / spouse line), kinship (blood relatives / married relatives), generational difference (previous generation / same generation / next generation), and confirmation status (confirmed / preset / pending confirmation / conflict).

[0052] S3024: Determine the relationship branch. The system determines whether the target person belongs to the direct line, paternal collateral branch, maternal collateral branch, spousal line, or descendant branch based on the key nodes in the path.

[0053] S3025: To determine whether someone is a blood relative or an in-law, the system checks if the path contains an edge with the "pair" relationship. If the path contains "pair" and the target person is not the individual's spouse, the in-law title is determined based on the location of the "pair".

[0054] S3026: Match title mapping table. The system matches the following core title mapping table based on the relation path encoding and auxiliary attributes, and outputs a unified standard title.

[0055] Example of the core appellation mapping representation: Relational path encoding auxiliary attributes Output standard name father Male, previous generation, direct descendant Father mother Female, previous generation, direct descendant Mother Mother-Father Maternal line, second generation, male maternal grandfather Mother-Mother Maternal line, second generation, female maternal grandmother Mother-Survivor, Male Maternal collateral branch, previous generation uncle Mother-sibling, female Maternal collateral branch, previous generation aunt Mother-Sperm-Child-Match Maternal collateral branches, contemporaries, and relatives by marriage Cousin's wife / brother's wife The actual system can expand more mapping entries according to regional language habits, but the unified standard name shall prevail when it is displayed to the outside world.

[0056] S3027: Output the superior title when information is insufficient. When the system lacks information such as gender, age, date of birth or confirmation status, output the superior standard title (such as "brother", "sister", "cousin", "spouse father" etc.). After subsequent information is supplemented, the title will be recalculated and updated to a more specific title.

[0057] S303: The system limits the kinship chain accessible to ordinary users to within four degrees. When receiving requests for Q&A, push notifications, recommendations, album categorization, or title queries, first determine the scenario type corresponding to the current task.

[0058] S304: The system distinguishes between user-initiated requests and system-triggered services based on the scenario type.

[0059] When users proactively raise questions related to family relationships, gift-giving, communication, holiday greetings, or expressions of affection, such as "What would be a suitable gift for my mother's birthday?" or "How should I communicate with my uncle?", these questions are categorized under the AI-powered emotional question-and-answer scenario. When the system actively identifies family events such as birthdays, holidays, anniversaries, weddings, birthday celebrations, family gatherings, elder care, and family interactions, it categorizes them into intelligent push scenarios.

[0060] S305: For AI-powered emotional question-and-answer scenarios, the system generates answers, communication suggestions, or action suggestions that fit the family relationship based on the relationship path between the user and the target relatives, standard titles, and desensitized map tags.

[0061] S306: For intelligent push scenarios, the system executes the following sub-steps: S3061: Family Incident Identification – Identifying incident types and determining key individuals involved in the incident; S3062: Push target selection - Based on the family gene chain, calculate the kinship path, degree of kinship, standard title and family role between the core person of the event and all related persons, and select first-degree, second-degree or other push targets that meet the scenario; S3063: Desensitization Map Tag Retrieval - Retrieve desensitization tags of key figures in the event, the target audience, and related relatives within a limited range of kinship degrees, including interests, consumption range, size, color preference, taboos, emotional sensitivities, historical gift feedback, historical interaction behavior, family roles, and kinship strength; S3064: Emotional Message Generation – Generate differentiated emotional reminder messages based on the relationship path between the push recipient and the key figures in the event, standard titles, family roles, and scenario types; S3065: Product or service matching - Based on event type, target person's preferences, target person's consumption habits, historical gift feedback, other relatives' gift-giving records, and product or service tags, match at least one product, service, virtual item, blessing, small red envelope, or local life service entry. S3066: Push and Transaction Entry Generation – Combine emotional language with product or service entry points to generate intelligent push content and send it to the corresponding user; after clicking, the user can enter the product details page, shopping cart, payment page, service appointment page, or virtual gift sending page.

[0062] S307: For family album categorization or standard title query scenarios, the system directly outputs the categorization results or title results. The de-identified tags do not directly expose sensitive personal fields to ordinary users.

[0063] Through the above steps, within the limited range of kinship degrees, relevant nodes are selected according to the application scenario, and their desensitized graph tags are called to execute intelligent services for the family scenario.

[0064] The advantages of this invention are as follows: It constructs a dynamic family gene chain centered on the individual using recursive expansion and shortest path re-judgment mechanisms, avoiding incorrect categorization of duplicate nodes; through preset kinship nodes, a matching confidence model based on multi-source evidence fusion, and a mutual confirmation mechanism, it achieves secure identification and automatic cross-network merging of unregistered relatives, eliminating the risk of incorrect binding and conflict; based on this, it combines the personal knowledge graph and standard title calculation for each individual node, and only calls de-identified tags to execute intelligent services within a four-degree range. This provides a real, continuous, and verifiable kinship background for scenarios such as intelligent question answering, personalized push notifications, family event recommendations, album categorization, and title queries while fully protecting user privacy, significantly improving the accuracy, contextual relevance, and usability of intelligent services in family scenarios.

[0065] Example 2 like Figure 2 As shown, this embodiment describes the construction of family gene chains and the method for determining kinship degree.

[0066] First, the system creates a core node for the user and assigns it a unique account identifier, node identifier, and initial relationship attributes. The user's own node is defined as a zero-degree node.

[0067] Users add their first-degree relatives nodes through the relationship topology graph interactive interface. The first-degree relatives nodes include father, mother, spouse, sibling, and child nodes, and the shortest path length between these nodes and the user node is 1.

[0068] For any person node in any degree layer, the system continues to use its father, mother, spouse, siblings, and children as entry points for expanding relationships, forming a candidate node set. After the candidate node set is formed, the system first performs duplicate person identification and node merging, and then calculates the shortest relationship path length of each merged person node relative to its own node. Only when the shortest relationship path length of a person node relative to its own node is equal to the target degree is the person node classified into the target degree layer; if the shortest path length is shorter, its original lower degree classification is maintained.

[0069] For example, User A sets themselves as the zero-degree node and adds their father, mother, spouse, sibling, and children as first-degree relatives. Then, they expand their list of relatives around their mother node, resulting in a candidate node set including maternal grandparents, uncles or aunts, father, and themselves or their siblings. After duplicate identification, node merging, and shortest path re-evaluation, maternal grandparents, uncles or aunts are classified as second-degree relatives; the mother's spouse is identified as the existing first-degree relative "father"; and the mother's children are identified as the zero-degree node "themself" or the existing first-degree relative "siblings." This demonstrates that candidate nodes do not directly determine degree; degree is determined by the shortest path length relative to the user's own node.

[0070] Through this embodiment, the system can dynamically construct a family gene chain centered on an individual and accurately determine the degree of kinship of each relative.

[0071] Example 3 This embodiment divides duplicate person identification into two categories: topologically deterministic duplicates and identity-deterministic duplicates, and adopts different judgment strategies for each.

[0072] Category 1: Topologically deterministic repetition (within the same relation graph) When user A has already added the node "Father Zhang" (a confirmed first-degree father-son relationship), and then adds "Father Zhang" around his "sibling" node (e.g., younger brother, who has been confirmed as a full sibling), the system recognizes that both nodes point to the same real person. Since both paths originate from edges with confirmed relationships, and the sibling relationship is clearly defined as full siblings, the system directly determines it as a topologically deterministic duplicate, without using the phone number as a necessary condition.

[0073] The system performs node merging: retaining the two relationship edges "self-father" and "sibling-father", merging the two nodes into a single main node, recalculating the shortest path length (still 1), and maintaining the one-degree-of-family categorization. This process does not require confirmation from either party; only user confirmation of the merge is needed (or the system will automatically merge and prompt the user).

[0074] Category 2: Duplicate claims with uncertain identities (across users, pre-defined nodes) User B pre-added a node for "Mother Li," and User C pre-added a node for "Li" as her sister. Li herself has not yet registered. Later, Li registers using her mobile phone number, and the system searches for matching pre-defined family member nodes. At this point, it is necessary to calculate the matching confidence using mobile phone number matching, invitation links, basic information consistency, relationship topology consistency, and a mutual confirmation mechanism.

[0075] If only the names match but there is no strong evidence such as a phone number, the C value will be low (e.g., 50.1), and the system will not prompt for automatic merging, but will only retain the candidate status in the background. Merging will only be performed when the phone number matches or the registration is made through an invitation link, and the basic information matches the relationship topology with a high degree of matching (C≥85), and after confirmation by both the person and the other party in the relationship.

[0076] In this embodiment, the system can correctly deduplicate without relying on mobile phone numbers in scenarios with clear topology, and adopts a strict confidence model in scenarios with uncertain identities, thus balancing efficiency and security.

[0077] Example 4 like Figure 4 As shown, this embodiment describes the complete process of setting up family member nodes, registering and claiming, and confirming the merger by both parties.

[0078] User A presets and adds the node "Mother Li", recording its relationship position, basic identity information and pending confirmation status. User B presets and adds "Li" as his sister. Li herself has not registered yet.

[0079] Subsequently, Li completes the registration using her mobile phone number, and the system retrieves preset family nodes that highly match the newly registered user. The system performs rough matching based on primary evidence (mobile phone number, real-name information), and then performs fine matching in combination with secondary evidence (name and birthday, relationship path consistency, cross-identification by multiple persons), and determines that the registered account highly matches both the two preset nodes of User A and User B.

[0080] The system displays candidate relationship prompts to Li, asking whether she is A's mother and B's sister. After Li confirms, the system initiates relationship confirmation to A and B respectively. After both parties complete the confirmation, the system merges Li's real registered account node with User A's preset mother node and User B's preset sister node, and automatically closes the relationship path formed between User A and User B due to sharing the same family member.

[0081] If Li refuses to claim or any party fails to complete the confirmation, the nodes remain in a separated state and no merging is performed.

[0082] Through this embodiment, the system realizes a security closed loop of "establishing a chain first, then claiming, and then merging", and supports automatic closure of cross-user relationship networks.

[0083] Fifth Embodiment As shown Figure 5 , this embodiment describes in detail the calculation method of standard appellations, which specifically includes shortest path acquisition, relationship path encoding, auxiliary attribute reading, branch determination, in-law judgment, mapping table matching, and superordinate appellation output when information is insufficient.

[0084] Specifically, the system takes the self node as the starting point and the target person node as the end point, and executes the following sub-steps: Step 1: Acquire the shortest relationship path; The system starts from the self node and calculates the shortest relationship path to the target node. If there are multiple reachable paths, the shortest path is preferentially adopted; if the paths are of equal length, the selection is made according to the priority of confirmed paths, blood relationship paths, and direct relationship paths.

[0085] Step 2: Generate relationship path encoding; Convert the relationships of father, mother, spouse, sibling, and child in the path into single-character Chinese encodings respectively: "father, mother, spouse, sibling, child". For example, the path "Self → Mother → Sibling → Child → Spouse" is encoded as "mother-sibling-child-spouse".

[0086] Step 3: Read auxiliary attributes; Read the gender (male / female / unknown), age (older / younger / unknown), lineage (paternal / maternal / spouse line), kinship type (blood relatives / married relatives), generational difference (previous generation / same generation / next generation), and confirmation status of the target node and key nodes in the path.

[0087] Step 4: Determine relationship branches and assess kinship; Determine whether the target belongs to the direct line, paternal collateral line, maternal collateral line, or spousal line based on the key nodes in the path; if the path contains "spousal" and the target is not the individual's spouse, then treat it as a spousal line.

[0088] Step 5: Match the title mapping table; By combining path encoding with auxiliary attributes, a pre-defined core title mapping table is searched, and a unified standard title is output.

[0089] Step 6: Output the superior title if there is insufficient information; If gender, age, or confirmed status is missing, output the superior title (such as "brother", "cousin", "spouse / father"), and recalculate and update it after the information is supplemented.

[0090] Specific examples: User A queries the title "wife of her mother's sister's son". The system executes the following: Shortest path: Self → Mother (first degree) → Mother's sister (second degree) → Sister's son (third degree) → Son's wife (fourth degree).

[0091] Path encoding: Mother → Cell (the mother's sister is considered her sibling) → Child (the sibling's offspring) → Gamete (the child's spouse). The encoding sequence is: Mother-Cell-Child-Gamete.

[0092] Auxiliary attributes: Target gender: female ("wife" is female); Age relationship: It is necessary to determine the age relationship between "the mother's sister's son" and the individual. In this example, the cousin may be older or younger, but since the system already has the birth date, it is determined to be younger. Lineage: Maternal collateral branch (because it extends from the mother's side); Kinship type: Affinal (path includes "marriage").

[0093] Branch identification: maternal collateral branches, same-generation in-laws.

[0094] Mapping table matching: Look up the code mother-sibling-child-mate, with the auxiliary attribute "male cousin and young", and output the standard title "cousin's wife".

[0095] If the system lacks the cousin's birth date and cannot determine the seniority, it outputs the superior title "cousin's spouse"; after the user provides the birth date later, the system recalculates and updates it to "cousin's wife" or "cousin's sister-in-law".

[0096] Error correction method: Users cannot directly edit the titles. If a title is incorrect, it indicates an error in the relational topology (e.g., incorrectly adding "mother's sister" as "father's sister"). Users must delete the incorrect node and re-add it in the correct position. The system will automatically recalculate the title to ensure consistency with the topology.

[0097] Through this embodiment, the system transforms title calculation from an abstract description into an implementable process of path encoding, attribute reading, and mapping table matching, fully supporting the output of superior titles when information is missing and subsequent dynamic updates.

[0098] Example 6 like Figure 6 As shown, this embodiment describes an intelligent push method based on family gene chains and personal knowledge graphs, specifically including the entire process of the system actively identifying family events, filtering push targets, generating differentiated emotional messages, and matching product / service entry points.

[0099] When the system detects that user Zhang San's son is about to have his birthday, it performs the following operations: Step 1: Family Event Identification: The system determines the event type as "children's birthday" and the key person in the event is Zhang San's son.

[0100] Step 2: Target Selection: The system calculates the kinship path between the key figure in the event (the son) and all relevant relatives based on the family gene chain. The result is: Zhang San: Father (once); Zhang San's spouse: Mother (once); Zhang San's mother: Grandmother (second time); Zhang San's father: grandfather (second generation); Zhang San's maternal grandparents: maternal grandmother and maternal grandfather (twice).

[0101] Step 3: Desensitization Chart Tag Retrieval: The system retrieves the son's chart preferences (e.g., "likes Ultraman"), size, and taboos (e.g., "mango allergy"); retrieves Zhang San's chart budget preferences ("under 500 yuan") and purchasing habits ("habitual JD.com shopping"); retrieves Zhang San's spouse's chart recent interests ("children's educational toys"); and retrieves other relatives' historical gift-giving records (to avoid duplication).

[0102] Step 4: Generating Emotional Scripts A message was sent to Zhang San: "Dear Dad, your son Zhang's birthday is coming soon (in 3 days). As a father, your company is the best gift. We have selected the following gifts suitable for your son. Click to view." Send a message to Zhang San's spouse: "Dear Mom, Zhang's birthday is coming soon. You are always so thoughtful, so we recommend some educational toys and picture books for children. You can also send a small red envelope to express your feelings." Send a message to Zhang San's mother (grandmother): "Grandma, your grandson Xiao Zhang is having a birthday party! Would you like to send him a blessing? We have prepared a gift and a voice blessing card suitable for your grandson." Step 5: Matching goods or services: For Zhang San: Match children's themed toys and children's sports shoes (price range 300-500 yuan), with e-commerce platform purchase links attached; For Zhang San's spouse: Match with children's educational building blocks and picture book sets, plus e-commerce coupons; For grandmothers: "Matching voice blessing card making service, children's story machine, with local merchant coupons."

[0103] Step 6: Push Notifications and Transaction Entry Generation: The system combines the aforementioned emotional language with product / service entry points and pushes them to users such as Zhang San, Zhang San's spouse, and Zhang San's mother. Users can then click to access the product details page, payment page, or greeting card editing page.

[0104] In this embodiment, the system transforms family events into proactive emotional reminders and transaction opportunities, which not only enhances the emotional interaction between family members, but also enables precise delivery of goods and services based on real family relationships.

[0105] Example 7 This embodiment describes a differentiated push method based on family gene chains and personal knowledge graphs.

[0106] When the system recognizes a family event, such as user A's son's birthday, the system generates differentiated push content for different service recipients.

[0107] For User A, the system generates push notifications for "Children's Birthday Reminders + Spouse Care Reminders". For User A's mother, the system generates push notifications for "Grandchildren's Birthday Reminders + Daughter-in-law Care Reminders". For User A's spouse, the system generates push notifications for "Children's Birthday Reminders + Family Preparations Reminders".

[0108] The content pushed to different recipients is determined by their relationship path with the target person of the event, their personal graph tags, and their family roles.

[0109] Through this embodiment, the system can provide personalized, family-relationship-appropriate intelligent push notifications for different roles in family events.

[0110] Example 8 This embodiment describes an automatic classification method for family photo albums based on family gene chains.

[0111] After users upload photos of family gatherings, the system categorizes the photos into personal albums, family albums, or clan albums based on the event tags corresponding to the photos, the results of people recognition, and the relationship paths in the family gene chain.

[0112] For example, for photos of maternal family gatherings, the system can categorize the photos into maternal branch albums based on the shortest relationship path. For shared family events (such as family portraits), the system simultaneously builds multidimensional indexes by people, by event, and by relationship branches.

[0113] Regular users can only view the corresponding de-identified association results within a four-dimensional range.

[0114] Through this embodiment, the system realizes automatic and multi-dimensional classification of photos, improving the management efficiency of family digital assets.

[0115] Example 9 This embodiment describes an AI question-answering optimization method based on family gene chains and personal knowledge graphs.

[0116] A user asked, "How should I communicate with my uncle about something?" The system first identifies the "uncle" as a second-degree maternal relative. Then, it retrieves the user's personal family tree, the uncle's family tree, and relevant family background tags. These tags include information such as communication style, emotional expression, frequency of past interactions, and family power dynamics. Based on the anonymized tag data, the system outputs communication solutions, precautions, and wording that are more relevant to real family relationships.

[0117] All data accessed uses de-identified tags as the primary input, and does not directly expose sensitive personal fields to ordinary users.

[0118] Through this embodiment, the system can improve the contextual relevance of question-and-answer sessions while effectively controlling the scope of exposure of sensitive information.

[0119] Example 10 This embodiment provides a matching confidence calculation model for identifying the same person between preset family member nodes and real registered user nodes, as well as a node merging and automatic relationship network closure method based on the model.

[0120] I. Model Definition and Core Objects This model is used to determine whether a preset relative node and a real registered user node point to the same real person. The model does not require mandatory real-name registration; it primarily relies on mobile phone registration information, preset node locations, basic information, relationship topology, and confirmation results from both parties to make the determination.

[0121] Define the following core objects: P: Preset family node, created by a registered user when a family member has not registered, and is in a pending confirmation state; R: Real registered user node, a system user node formed after a user registers using their mobile phone number; U: Preset node creator, i.e., the registered user who added the preset relative; C: Match confidence level, ranging from 0 to 100.

[0122] II. Formula for Calculating Matching Confidence The matching confidence level C is calculated using the following formula: C=100×(0.45I+0.20B+0.35T)−P in: I represents the score for evidence of contact information and registration source; B is the basic information consistency score; T represents the relation topology consistency score; P represents the conflict penalty score.

[0123] If the calculation result is less than 0, it is treated as 0; if it is greater than 100, it is treated as 100.

[0124] III. Scoring Rules for Each Item 3.1 Contact information and evidence of registration source I (weight 45%): The default node mobile phone number is the same as the registered mobile phone number: I=1.00; Registered via a family member's invitation link, and the link is bound to the preset node: I=0.90; The last four digits of the phone number are the same, and the name is also the same: I=0.65; Name only, no phone number or proof of registration source: I=0.00; If the phone numbers are clearly inconsistent and there is no other strong evidence, proceed to conflict assessment.

[0125] Real-name information is an optional verification item. If a user voluntarily provides real-name information and the information matches, it can be used as a strong verification basis; if there is a clear conflict in the real-name information, automatic merging will be prohibited.

[0126] 3.2 Basic Information Consistency B (Weight 20%): B = 0.50B1 + 0.35B2 + 0.15B3 in: B1 represents the name consistency score; B2 is the consistency score for birth date or age group; B3 represents the gender consistency score.

[0127] Value rules: 1.00 for complete consistency, 0.60 for partial consistency, 0.30 for missing information, and 0 for explicit conflict, which then enters the conflict judgment process.

[0128] 3.3 Relationship topological consistency T (weight 35%): T = 0.40T1 + 0.30T2 + 0.20T3 + 0.10T4 in: T1 is the relationship role consistency score: it determines whether the real registered user is in the relationship position corresponding to the preset node, such as whether it is indeed "mother" or "child"; T2 is the shortest path degree consistency score: after temporary merging, the shortest path is recalculated to determine whether the degree of the person is correct; T3 is the kinship topology legality score: checking for issues such as conflicts between direct blood relatives and spouses, reversed father-son relationships, and unreasonable generational relationships. T4 is the relational closure gain score: it determines whether a reasonable relational network can be formed after node merging.

[0129] 3.4 Conflict Penalty Points (P): The system will execute the corresponding conflict penalty when the following situations occur: The user explicitly refused to claim the item: P=100, merging is prohibited; Conflicts in voluntary real-name information: P=100, merging is prohibited; Confirmed to be bound to another real person: P=100, merging is prohibited; Conflict between lineal relatives and spouses: P=100, merger is prohibited; Significant conflict between gender and father / mother relationship: P=60 or merging is prohibited; A clear conflict exists between age and generational relationships: P=30 to 60; The merged nomenclature path cannot be calculated: P=50; The phone numbers are clearly inconsistent and there is no other strong evidence: P=40.

[0130] Once P=100, the system no longer considers the confidence score and directly prohibits merging.

[0131] IV. Candidate Generation and Merging Rules The system performs different processing based on the value of the confidence level C: C≥85: Generate a strong candidate relationship and initiate relationship confirmation to both parties; 70≤C<85: Generate weak candidate relationships, prompt only or require manual / user confirmation; C<70: No merge prompt; There is a hard conflict: Merging is prohibited.

[0132] The final merger must meet all of the following conditions simultaneously: C has reached the merging threshold; The identified relative has been confirmed by the individual. Confirmation by the other party to the relationship; There is no hard conflict; The shortest path metric is correct after merging.

[0133] V. Merging of two registered users after confirming their relationship Once two registered users have confirmed their relationship, the system does not directly merge all the preset family nodes of both parties. Instead, it uses the two confirmed user nodes as the network merging anchor points, generates candidate merging relationships around the existing preset family nodes of both parties, and then calculates the matching confidence level one by one.

[0134] The specific steps are as follows: Step 1: Merge or confirm the primary relationship anchor between the two registered users; Step 2: Based on the confirmed relationship, deduce the candidate corresponding relationships of the pre-set relatives of both parties; Step 3: For each group of candidate nodes, calculate C=100×(0.45I+0.20B+0.35T)−P according to the above formula; Step 4: For candidate nodes that have reached the merging threshold and have no conflicts, initiate confirmation or execute the merge; for low-confidence candidate nodes, only retain the candidate association status.

[0135] For example, user A confirms that user B is their mother. Then, candidate correspondences are generated between A's preset "maternal grandfather," "maternal grandmother," "uncle / aunt," "sibling," and "father," and B's preset "father," "mother," "sibling," "children," and "spouse." Among these, "A's maternal grandfather" and "B's father" are high-probability candidates and require further confidence calculation; "A's father" and "B's spouse" are only considered as candidates and are not directly merged to account for real-world situations such as divorce, remarriage, and step-relatives.

[0136] The advantages of this invention are: it can accurately determine whether a preset relative node and a real registered user node are the same real person by fusing multiple evidences based on mobile phone number, basic information, and relationship topology without requiring mandatory real-name registration. The model effectively avoids erroneous merging through weight allocation, conflict penalties, and a two-party confirmation mechanism. For cross-user relationship networks, candidate merging relationships are generated using confirmed relationships as anchor points and verified one by one, achieving secure and controllable automatic closure of the family gene chain while retaining the flexibility to handle complex family relationships (such as divorce and remarriage).

[0137] Example 11 like Figure 7 As shown, a family gene chain node recognition and intelligent service system includes: M1 Family Chain Construction Module: Centered on the user's own node, it forms a dynamically scalable family gene chain through recursive expansion, candidate set generation, duplicate person identification and node merging, and shortest path grading. M2 Node Confidence Calculation Module: Based on a multi-source evidence fusion matching confidence model (formula C=100×(0.45I+0.20B+0.35T)−P), calculates the matching confidence between preset nodes and registered user nodes; M3 Relationship Confirmation and Node Merging Module: Performs mutual confirmation, node merging, retains all source relationship edges, and refreshes the shortest path and standard names; M4 Desensitized Tag Retrieval Module: Extracts desensitized map tags within a limited range of fourth-degree kinship, based on the application scenario; M5 Smart Service Module: Performs intelligent services for family scenarios such as AI-powered emotional Q&A, intelligent push notifications, family event recommendations, photo album categorization, and standard title lookup; M6 storage module: Stores multidimensional semantic graph information for each character node, including its source, version, credibility parameters, relationship edges, and confirmation status.

[0138] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for node adoption, relationship merging, and intelligent services based on family gene chains, characterized in that, include: Centered on the user's own node, a dynamic and scalable family gene chain is formed by recursively expanding and performing shortest path-based merging and degree re-evaluation on duplicate nodes generated during the expansion. By pre-setting family node, calculating matching confidence based on multi-source evidence fusion, and confirming the relationship between the two parties, the newly registered user node is merged with its corresponding pre-set family node; Within a limited range of kinship degrees, relevant nodes are selected based on the application scenario, and their anonymized graph tags are invoked to execute intelligent services for the family scenario.

2. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, The construction of the family gene chain specifically includes: taking the user's own node as the zero-degree node, recursively expanding the relationship edges around five types of relationships: father, mother, spouse, sibling, and children, to form a candidate node set; performing duplicate person identification and node merging on the candidate node set, and calculating the shortest relationship path length of each merged node relative to the user's own node; only when the shortest path length is equal to the target degree is the node classified into that degree level, otherwise its original lower degree classification is maintained.

3. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, The preset relative node includes: when the target relative has not yet registered, the preset relative node is created by the registered user. This node occupies a certain position in the relationship graph and records the relationship type, basic identity information and pending confirmation status; when the target relative subsequently registers through a mobile phone number to form a new registered user node, the system retrieves the matching preset relative node and generates a matching confidence score.

4. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 3, characterized in that, The matching confidence score based on multi-source evidence fusion is calculated using the following formula: C=100×(0.45I+0.20B+0.35T)−P Wherein, I is the score for contact information and registration source evidence, B is the basic information consistency score, T is the relationship topology consistency score, and P is the conflict penalty score; the contact information and registration source evidence includes consistent mobile phone number, consistent de-identified mobile phone number characteristics, or invitation link binding; the basic information consistency includes name, date of birth or age group, and gender; the relationship topology consistency includes relationship role consistency, shortest path degree consistency, kinship topology legality, and relationship closure gain; the conflict penalty score is taken from 0 to 100 based on user rejection, real name conflict, conflict between direct blood relatives and spouses, gender and relationship conflict, age and generational conflict, and inability to calculate the title path, etc., where merging is prohibited when P=100.

5. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 4, characterized in that, The conditions for confirmation and node merging between the two parties in the relationship are as follows: when the matching confidence C≥85, the system initiates confirmation between the identified relative and the other party in the relationship; the system will only merge the newly registered user node with the original preset relative node when the identified relative confirms, the other party in the relationship confirms, there is no hard conflict, and the shortest path gradation after merging is correct; when C<70, no merging is prompted.

6. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, It also includes the merging process after two registered users confirm their relationship: After two registered users complete the relationship confirmation, the system uses the two newly registered user nodes as merging anchors, generates candidate merging relationships around the existing preset family nodes of both parties, and calculates the confidence of each candidate group one by one according to the relationship mapping rules and the matching confidence model. For candidate nodes that have reached the merging threshold and have no conflicts, initiate confirmation or execute the merging; for low-confidence candidate nodes, only retain the candidate association status.

7. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, It also includes binding a personal knowledge graph to each node in the family gene chain: the personal knowledge graph contains multiple pieces of information including basic attributes, interests and preferences, behavioral habits, health taboos, consumption style, family roles, relationship memories, emotional triggers, and behavioral feedback; it records the data source, timestamp, version number, and credibility parameter for each graph field; it assigns different credibility to data from different sources, retains historical versions, and performs weight decay processing on new and old preferences; when information from different sources conflicts, it prioritizes the version with higher credibility and more recent time.

8. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, The intelligent service for family scenarios also includes standard title query: obtain the shortest relationship path from the user node to the target node, read the relationship type, generational direction, paternal or maternal attribute, blood or in-law attribute and seniority attribute of each relationship edge on the path, and output a unified standard title according to the preset title mapping rules. Users cannot directly edit the title results; they must modify them indirectly by correcting the relational topology.

9. The method for node adoption, relationship merging, and intelligent services based on family gene chains according to claim 1, characterized in that, The smart home services also include at least one of the following: AI-powered emotional Q&A: When users initiate questions about family relationships, gift-giving, communication, holiday greetings, or expressions of affection, the system generates answers, communication suggestions, or action suggestions tailored to family relationships based on the user's relationship path with the target relatives, standard titles, and desensitized map tags. Intelligent push notification: When the system actively identifies family events such as birthdays, holidays, anniversaries, weddings, birthday celebrations, family gatherings, elder care, or kinship interactions, it determines the relationship path, standard titles, degree of kinship, and family roles between the key figures of the event and the push recipients based on the family gene chain. It then calls the desensitized map tags of the key figures of the event, the push recipients, and related relatives to generate differentiated push content with emotional expression. Finally, it matches at least one product, service, virtual item, blessing, small red envelope, local life service, or online mall entrance and pushes it to the corresponding user. Family album categorization: Based on the event tags of the photos and the relationship path of the identified person nodes in the family gene chain, the photos are automatically categorized into the corresponding relationship branch albums; Standard title query: Obtain the shortest relationship path from the user node to the target node, read the relationship type, generation direction, paternal or maternal lineage, blood or in-laws, and seniority attributes of each relationship edge on the path, and output a unified standard title according to the preset title mapping rules.

10. A family gene chain node recognition and intelligent service system, characterized in that, The method for node adoption, relationship merging, and intelligent services based on family gene chains according to any one of claims 1 to 9 includes: The family chain construction module takes the user's own node as the center, and forms a dynamic and scalable family gene chain by recursively expanding and performing shortest path-based merging and degree re-judgment on duplicate nodes generated during the expansion. The node and confidence module merges the newly registered user node with its corresponding preset family node through preset family node, matching confidence calculation based on multi-source evidence fusion, and confirmation from both parties to the relationship; The merging and confirmation module, within a limited range of kinship degrees, filters relevant nodes based on the application scenario and calls their desensitized graph tags to execute intelligent services for the family scenario; The desensitization module is used to execute intelligent family scene services by calling desensitized map tags according to the scenario within a limited range of relatives' degrees; The storage module is used to store the multidimensional semantic graph information of each character node, as well as its source, version, and credibility parameters.

Images