Network node and method for ontology mapping in a communication network

Abstract

A method performed by a network node is provided The method is for mapping ontologies in a communications network. The network node obtains (201) a first ontology, Ontology 1, representing data of a first set of data. The network node obtains (201) a second ontology, Ontology 2, representing data of a second set of data. The network node collects (202) a number of Ontology 1 nodes from the data of the first set of data. Each respective Ontology 1 node comprises data from the first set of data. The network node collects (202) a number of Ontology 2 nodes from the data of the second set of data. Each respective Ontology 2 nodes comprises data from the second set of data. For each respective node out of said Ontology 1 nodes and said Ontology 2 nodes, the network node generates (203) a node description based on a content of the data in the node and embeds (204) the node description, capturing a semantic context associated with the node. The network node determines (205) similarity scores between the embedded Ontology 1 node descriptions and embedded Ontology 2 node descriptions. The network node establishes (206) intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2. This is based on the Ontology 1 node descriptions and Ontology 2 node descriptions. The network node then maps (207) the ontologies by mapping Ontology 1 nodes with Ontology 2 nodes based on the determined similarity scores and the established intra-ontology causality.

Description

[0001] NETWORK NODE AND METHOD FOR ONTOLOGY MAPPING IN A COMMUNICATION NETWORK

[0002] TECHNICAL FIELD

[0003] Embodiments herein relate to a network node and methods therein. In some aspects, embodiments herein relate to mapping ontologies in a communications network.

[0004] BACKGROUND

[0005] In a typical wireless communication network, wireless devices, also known as wireless communication devices, mobile stations, stations (STA) and / or User Equipment (UE), communicate via a Wide Area Network or a Local Area Network such as a Wi-Fi network or a cellular network comprising a Radio Access Network (RAN) part and a Core Network (CN) part. The RAN covers a geographical area which is divided into service areas or cell areas, which may also be referred to as a beam or a beam group, with each service area or cell area being served by a radio network node such as a radio access node e.g., a Wi-Fi access point, a Base Station (BS) or a radio base station (RBS), which in some networks may also be denoted, for example, a Base Station (BS), a NodeB, eNodeB (eNB), or gNodeB (gNB) as denoted in Fifth Generation (5G) telecommunications. A service area or cell area is a geographical area where radio coverage is provided by the radio network node. The radio network node communicates over an air interface operating on a radio frequency with the wireless devices within the range of the radio network node.

[0006] 3rd Generation Partnership Project (3GPP) is the standardization body for specifying the standards for the cellular system evolution, e.g., including 3G, Fourth Generation (4G), 5G and the future evolutions. Specifications for Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Packet System (EPS) have been completed within the 3GPP. In 4G, EPS is core network and E-UTRA is radio access network. In 5G, 5G Core (5GC) is core network, NR is radio access network. As a continued network evolution, the new release of 3GPP specifies a 5G network also referred to as 5G New Radio (NR) and 5GC.

[0007] Frequency bands for 5G NR are being separated into two different frequency ranges, Frequency Range 1 (FR1) and Frequency Range 2 (FR2). FR1 comprises sub-6 GHz frequency bands. Some of these bands are bands traditionally used by legacy standards but have been extended to cover potential new spectrum offerings from 410 MHz to 7125 MHz. FR2 comprises frequency bands from 24.25 GHz to 52.6 GHz. Bands in this millimeter wave range have shorter range but higher available bandwidth than bands in the FR1.

[0008] Multi-antenna techniques may significantly increase the data rates and reliability of a wireless communication system. For a wireless connection between a single user, such as UE, and a base station (BS), the performance is in particular improved if both the transmitter and the receiver are equipped with multiple antennas, which results in a Multiple-Input Multiple-Output (MIMO) communication channel. This may be referred to as Single-User (SU)-MIMO. In the scenario where MIMO techniques is used for the wireless connection between multiple users and the base station, MIMO enables the users to communicate with the base station simultaneously using the same time-frequency resources by spatially separating the users, which increases further the cell capacity. This may be referred to as Multi-User (MU)-MIMO. Note that MU-MIMO may benefit when each UE only has one antenna. The cell capacity can be increased linearly with respect to the number of antennas at the BS side. Due to that, more and more antennas are employed in BS. Such systems and / or related techniques are commonly referred to as massive MIMO.

[0009] In an era where information is abundant, a seamless integration and interpretation of data from different sources has become a critical issue across many fields. One possible solution to this issue is ontology mapping, which is the process of establishing connections between objects in different ontologies. An ontology is a formal representation of knowledge that defines concepts within a domain and relationships between those concepts, provide structured framework for organizing and interpreting data. It should be noted that the wordings “Concept”, “Entity”, “Node”, “Object” and “Field” as used herein have an equal or similar meaning and may be used interchangeably herein. Ontologies represent information through different features such as classes and instances their properties and axioms that govern the relationships between these classes. To process different ontologies, ontology mapping is crucial. Ontology mapping is a process of establishing correspondences between different entities or concepts in different ontologies. Ontology mapping plays a significant role in various domains and applications, contributing to knowledge integration, semantic interoperability, and enhanced data utilization. Typical application includes data migration and interoperability, decision support system, information integration, semantic web development, natural language processing, ontology merging, and knowledge graph creation. To apply and illustrate the idea of ontology mapping an example of a domain will be described below.

[0010] A migration route for moving subscriber configuration and business from legacy telecom products to new telecom products must always be developed by telecom operators before new products are released. While Telecom products must follow the TM Forum standard, this isn't always the case because most legacy products were on the market long before the TM standard was created. Due to this, telecom providers are required to develop a migration flow for each new product they migrate.

[0011] SUMMARY

[0012] As part of developing embodiments herein, the inventors identified some problems that first will be described.

[0013] It is simpler to transform any telecom product's configuration into a causal structure in order to take advantage of ontology and causal structure. Future product migration will be aided by a method according to embodiments herein, of constructing the telecom legacy and new product in a causal structure and creating a semantic match to map the components between the products, all without the need for human creation.

[0014] To significantly enhance the promise of ontology and ontology mapping for product migration, causal relationships of the objects present in the product ontologies need to be considered. Causality investigates a cause-and-effect relationship between variables where a change in one variable may affect another variable. In case variables, the objects are inside the ontologies.

[0015] Understanding the motivations underlying the interconnections among objects in ontology, is a significant step in understanding the ontology, it also gives the possibility of using causal analytics in downstream tasks which use the ontology, i.e. , causality enables ontology that illuminates the "why" behind data linkages. For many applications, this viewpoint may be very beneficial for predicting outcomes, uncovering hidden insights, and making well-informed decisions. Ontology integration is being investigated in several applications, particularly in the telecom domain.

[0016] Existing ontology and ontology mappings schemes are not utilizing the causal relationship between the objects present in the ontology. In a typical ontology, there may be associated data available with the objects and objects may have causal relation between them. Hence, knowing the causal relation between the entities in an ontology helps to better utilize these ontologies for downstream processing tasks and also helps in better interpretation of ontologies. However, existing ontology representation schemes are not representing the causal relationship between the objects. The causal structure associated with ontology enable one to perform causal analytics, causal inferencing, counterfactual analysis, and root cause analysis, on the objects and also give better interpretation of objects and model development.

[0017] BertMap is an ontology alignment system that uses synonyms in the ontologies to fine-tune the BERT model to find mappings. This model has achieved state-of-the-art accuracy in certain ontology mapping tasks, but it ignores scenarios in which the labels in the ontologies lack synonyms necessary to fine-tune BERT. BertMap represents one of the works that makes use of the semantic relationships between the objects to predict the mapping. By encouraging a GPT-3.5 model to attempt to map the two ontologies by prompting it into giving the correct mapping, another work He, Yuan, et al. "Exploring Large Language Models for Ontology Alignment." arXiv preprint arXiv:2309.07172 (2023), suggests using large language models for ontology alignment. Although the model shows potential, it falls short of BERTMap's performance.

[0018] There are also works such as Yang, Feng. "Variable weight semantic graph-based ontology mapping method." Expert Systems 36.5 (2019): e12337, which make use of both a graph structure inside ontologies as well as the semantic meaning between the ontologies to map different objects. However, methods like these do not make use of the underlying causal structure inside the ontologies to better the mapping and might lose key information regarding the data to improve the insights and reinforcement that causal discovery provides.

[0019] In general, telecom product migration at present performs ontology mapping which aided by manual labelling and synonyms. It is a compulsory process the equivalent parameters are to be labelled with the help of humans to perform the ontology mapping. According to embodiments herein, an automate process is provided by removing manual intervention. An intermediate process is identified to remove the manual intervention required during ontology mapping.

[0020] The existing ontology mapping techniques rely on the semantic meaning of object names, where semantically similar objects in two ontologies are mapped. However, in an ontology, there may be object names that lack semantic comprehensibility / meaning. It may be common in product related ontologies where object names may come in abbreviations, contains numbers, versions number / names, etc. Another scenario arises when object names in two ontologies appear semantically similar, but no actual relations exist. In these situations, existing mapping strategies are prone to failure to introduce additional challenge in this process.

[0021] Conventional automatic ontology mapping techniques utilize the semantic meaning of the names of objects to find the mapping. However, in certain cases, object names in ontology may not possess any semantic meaning, e.g., Abbreviations, product serial number, port name from the data collected, etc. In these situations, semantic based ontology mapping schemes will fail.

[0022] Existing ontology mapping schemes use the semantic meaning of objects names to perform the mapping. However, these mapping schemes may fail because of the lack of semantic meaning present in the names of the objects. Moreover, these methods are not using data associated with objects to find the mapping.

[0023] An object of embodiments herein is to improve the way of mapping ontologies in a communications network.

[0024] Embodiments herein provide an ontology representation which reveals the existing causal structure between the entities in the ontology and an automatic ontology mapping technique which exploits a causal structure to obtain a robust ontology mapping, even in situations where there is no semantic relation between objects in ontologies. According to embodiments herein, findings are presented that causal discovery may enable ontology mapping in a better way and which discovers a better mapping between the ontologies.

[0025] According to an aspect of embodiments herein, the object is achieved by a method performed by a network node. The method is for mapping ontologies in a communications network. The network node obtains a first ontology, Ontology 1 , representing data of a first set of data. The network node obtains a second ontology, Ontology 2, representing data of a second set of data. The network node collects a number of Ontology 1 nodes from the data of the first set of data. Each respective Ontology 1 node comprises data from the first set of data. The network node collects a number of Ontology 2 nodes from the data of the second set of data. Each respective Ontology 2 nodes comprises data from the second set of data. For each respective node out of said Ontology 1 nodes and said Ontology 2 nodes, the network node generates a node description based on a content of the data in the node and embeds the node description, capturing a semantic context associated with the node. The network node determines similarity scores between the embedded Ontology 1 node descriptions and embedded Ontology 2 node descriptions. The network node establishes intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2. This is based on the Ontology 1 node descriptions and Ontology 2 node descriptions. The network node then maps the ontologies by mapping Ontology 1 nodes with Ontology 2 nodes based on the determined similarity scores and the established intra-ontology causality.

[0026] According to another aspect of embodiments herein, the object is achieved by a network node. The network node is configured to map ontologies in a communications network. The network node further being configured to obtain a first ontology, Ontology 1 , adapted to represent data of a first set of data. The network node further being configured to obtain a second ontology, Ontology 2, adapted to represent data of a second set of data. The network node further being configured to collect a number of Ontology 1 nodes from the data of the first set of data. Each respective Ontology 1 node comprises data from the first set of data. The network node further being configured to collect a number of Ontology 2 nodes from the data of the second set of data. Each respective Ontology 2 nodes comprises data from the second set of data. The network node is further being configured to, for each respective node out of said Ontology 1 nodes and said Ontology 2 nodes generate a node description based on a content of the data in the node and embed the node description, capturing a semantic context associated with the node. The network node further being configured to determine similarity scores between the embedded Ontology 1 node descriptions and embedded Ontology 2 node descriptions. The network node further being configured to establish intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the Ontology 1 node descriptions and Ontology 2 node descriptions. The network node further being configured to map the ontologies by mapping Ontology 1 nodes with Ontology 2 nodes based on the determined similarity scores and the established intra-ontology causality.

[0027] Embodiments herein may provide one or more of the following advantages:

[0028] - Enable employing of causal analysis tools on an ontology to e.g., obtain causal inference, counterfactual reasoning and explainability.

[0029] - Better ontology maps when object names are not semantically meaningful.

[0030] - Yields better semantic representation. - The integration of causal discovery inside ontologies with the semantic based matching approach gives an enhanced final mapping that not only represent relationships between ontologies but also represents and makes use of the causal relationships inside ontologies for a more robust detailed and accurate mapping.

[0031] - A scalability of embodiments herein to handle complex ontologies autonomously and an accurate ontology mapping, from the input ontologies, with very minimal interference from the user.

[0032] - Enhanced data integration, comprehensive understanding, informed decisionmaking and improved risk assessment.

[0033] - A better query processing by performing end-to-end ontology mapping architecture.

[0034] - Any telecom query such as e.g., subscriber migration, ticket handling process, vehicular computing, etc., which gives more meaningful results comparative to the general approaches.

[0035] BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Examples of embodiments herein are described in more detail with reference to attached drawings in which:

[0037] Figure 1 is a schematic block diagram illustrating an example of embodiments herein. Figure 2 is a flowchart depicting an embodiment of a method in a network node.

[0038] Figure 3 is a flowchart illustrating an example embodiment of a method herein.

[0039] Figure 4 is a flowchart illustrating an example embodiment of a method herein.

[0040] Figure 5 is a schematic block diagram illustrating an example embodiment of a method. Figure 6 is a schematic diagram illustrating an example embodiment.

[0041] Figure 7 is a sequence diagram illustrating an example embodiment.

[0042] Figure 8 is a schematic diagram illustrating an example embodiment.

[0043] Figure 9 is a schematic diagram illustrating an example embodiment.

[0044] Figure 10 is a schematic diagram illustrating an example embodiment.

[0045] Figure 11 is a schematic block diagram illustrating embodiments of a network node. DETAILED DESCRIPTION

[0046] Example embodiments herein provide causality augmented ontology representation and mapping. This may among other things, be used for product migration activities related to telecom subscribers.

[0047] Example embodiments herein provide an ontology representation and an automatic ontology mapping based on causal structure of objects present in the ontology. A causal discovery framework is used to analyse and identify a causal relation between ontology nodes, also referred to as objects, in the given ontology from the associated data.

[0048] Example embodiments herein provide an ontology mapping scheme which goes beyond semantic similarity based mapping, by using the causal structure of ontology nodes present in the ontologies where a causal structure is estimated from the associated data by e.g., using an efficient causal discovery algorithm. According to some example embodiments herein, the ontology mapping technique employ a score-based approach in an embedding domain to find the inter ontology mapping where a causal graph presenting ontologies may be utilized to obtain the scores. The proposed formulation provides a meaningful semantic map with causal relations to enhance the ontology mapping.

[0049] To illustrate the usefulness of ontology mapping, use cases of ontology alignment procedures are considered on telecom product migration data. It is observed that the provided methodology enables to come up with a meaningful ontological map where causal relations are incorporated, which leads to a better understanding of ontologies and also helps the downstream tasks to perform causal analytics on the resulting ontology map.

[0050] The provided mapping may efficiently modify the knowledge representation to enhance support for data integration and decision-making, such as e.g. in a telecom product migration scenario with an old telecom product configuration to be migrated to the new upgraded telecom product.

[0051] Figure 1 is a schematic overview depicting a communications network 100 wherein embodiments herein may be implemented. Figure 1 depicts example embodiments of mapping ontologies in the communications network 100. The communications network 100 comprises ontologies such as a first ontology referred to as Ontology 1 , and a second ontology, referred to as Ontology 2. The Ontology 1 represents data of a first set of data e.g., related to a first technology. The Ontology 2, represents data of a second set of data e.g., related to a second technology. Ontology 1 comprises a number of Ontology 1 nodes 11, 12, 13. Each respective Ontology 1 node 11, 12, 13 comprises data from the first set of data. Ontology 2 comprises a number of Ontology 2 nodes 21, 22, 23. Each respective Ontology 2 nodes 21, 22, 23 comprises data from the second set of data.

[0052] Figure 1 shows a simple output provided by some example embodiments, of mapping ontologies in the communications network 100. Figure 1 depicts a schematic diagram showing ontology mapping of concepts in Ontology 1 with concepts in Ontology 2. The ontology mapping is based on intra-ontology causality of concepts in Ontology 1 and inter-ontology semantic similarity between concepts in Ontology 1 and concepts in Ontology 2.

[0053] As mentioned above, the wordings “Concept”, “Entity”, “Node”, “Object” and “Field” have an equal or similar meaning and may be used interchangeably herein.

[0054] A network node 110 operates in the communications network 100. The network node 110 may e.g. be a comprise a list of services which helps in business operation of any telecommunications network product. For example, a Service Data Point (SDP ) is one such network node for handling resource consumption and e.g., charging a product. The method may comprise one or more of the following actions performed by the network node 110.

[0055] The network node 110 obtains ontology nodes e.g., from two distinct ontologies, each representing a respective datasets. For example, the network node 110 obtains Ontology 1 nodes 11 , 12, 13, e.g. by collecting nodes from Ontology 1 , and Ontology 2 nodes 21, 22, 23, e.g. by collecting nodes from Ontology 2. Each respective Ontology 1 node 11 , 12, 13 comprises data from the first set of data, and each respective Ontology 2 nodes 21 , 22, 23 comprises data from the second set of data.

[0056] The network node 110 generates Node Descriptions for all Ontology 1 nodes 11, 12, 13 and Ontology 2 nodes 21 , 22, 23. E.g. by utilize ChatGPT or similar language models to create detailed descriptions for the Ontology 1 nodes 11 , 12, 13 and the Ontology 2 nodes 21 , 22, 23.

[0057] The network node 110 generates Node Embeddings for the Ontology 1 nodes 11, 12, 13 and the Ontology 2 nodes 21 , 22, 23. E.g. by using sentence BERT to create embeddings for node descriptions, capturing semantic context Sentence BERT when used herein in this context will e.g. try to map the similarity in the network elements in a network product represented as Ontology format. This similarity is e.g. measured in terms of inter ontology and / or intra ontology. The network node 110 calculate Similarity Scores, referred to as Inter-ontology Semantic Similarity in Figure 1. The network node 110 may e.g. employ cosine similarity measures to determine the similarity scores between the node descriptions using transformer models.

[0058] The network node 110 further determines, also referred to as discovers, IntraOntology Causality. The network node 110 may e.g. apply a PC algorithm to identify causal relationships within each Ontology.

[0059] The network node 110 may utilize the Intra-Ontology Causality and the similarity scores such as the Inter-ontology Semantic Similarity for mapping. The network node 110 may e.g. combine the Intra-Ontology Causality information and similarity scores to map nodes between ontologies.

[0060] In some embodiments, the network node 110 arranges for displaying the mapped ontologies to the concerned product level ontologies. This may e.g., be performed by presenting the final mapping results, showcasing the interconnected ontologies with semantic and causal insights.

[0061] Embodiments herein provide an ontology representation that incorporates causal relations, also referred to as intra-ontology causality of nodes in a given ontology. Example embodiments herein use data associated with the nodes to identify the causal structure such as the Intra-Ontology Causality by using an appropriate causal discovery algorithm. The framework according to some embodiments herein enables a better understand of the ontology and uses causal analytic tools on the given ontology thereby obtaining a causal inference, a counterfactual reasoning, an explainability and other benefits of causal analytics.

[0062] Some example embodiments herein provide an automatic ontology mapping technique e.g., based on the causal graph allows one to go beyond the semantic mapping scheme. The causal structure overcomes the limitation that arises in the existing technique semantic based technique due to lack of semantic meaning and semantic ambiguity of the objects name. It may also induce zero-touch automation in the provided process.

[0063] The algorithm according to some embodiments herein may be specifically observed in a scenario where a situation is handled where the objects of a given ontology do not have required necessary descriptions. Example embodiments herein e.g., uses power of Large Language Models (LLMs) to solve this by generating descriptions on it’s own. The methodology herein may e.g., handle the situation with power of generative Al. According to an approach of embodiments herein, it is believed that it’s tailor-made nature to take on problems such as product migration e.g. in a telecom domain in a scalable architecture. However, the approach may easily be extended to relevant tasks of any other domain executed through ontology mapping. Some applications and use cases will be discussed below.

[0064] A number of embodiments will now be described, some of which may be seen as alternatives, while some may be used in combination.

[0065] A method according to embodiments will first be described in a general way as seen from the view of the network node 110 together with Figure 2. This will be followed by examples and a more detailed description.

[0066] Figure 2 shows exemplary embodiments of a method performed by the network node 110. The method is for mapping ontologies in a communications network 100.

[0067] The method comprises the following actions, which actions may be taken in any suitable order. Optional actions are referred to as dashed boxes in Figure 2.

[0068] Ontology 2 represents data of the new system to be migrated to. There is then a need to find corelations between the legacy system and the new system. E.g., what is similar and what differs in the two ontologies, and what will be better after migration and what will be worse but can be handled when known before the migration. Therefore, an ontology mapping according to embodiments herein would be helpful in the migration process to unravel a worse situation a. Ontology 1 mat represent a first product which is to mapped to the second product during product migration.

[0069] Action 201. The network node 110 obtains a first ontology, referred to as Ontology 1, representing data of a first set of data. The first set of data 1 may e.g., be related to a first technology. The network node 110 further obtains a second ontology, referred to as Ontology 2, representing data of a second set of data. The second set of data may e.g., be related to a second technology. According to an example scenario, these two ontologies relate to ontologies that are to be mapped to each other later on. Action 202. The network node 110 collects a number of Ontology 1 nodes 11, 12, 13 from the data of the first set of data. Each respective Ontology 1 node 11 , 12, 13 comprises data from the first set of data.

[0070] The network node 110 further collects a number of Ontology 2 nodes 21 , 22, 23 from the data of the second set of data. Each respective Ontology 2 nodes 21 , 22, 23 comprises data from the second set of data.

[0071] Nodes when used herein may e.g. comprise a list of business processing services pertained to a particular business need. For example, charging as a product may help telecom charging services to charge any user level event like subscriptions, plans and usage etc.

[0072] According to the example scenario, it is nodes of the two different ontologies that are to be mapped.

[0073] Action 203. For each respective node out of said Ontology 1 node 11, 12, 13 and said Ontology 2 nodes 21 , 22, 23 the network node 110 generates a node description based on a content of the data in the node. This may be to describe the nodes which e.g. may fields of the data. For example, if migrating from one legacy product to a another product , e.g., in a telecom domain, the different fields that will be described may relate to data rate, beam forming, coverage ratio, etc.

[0074] Action 204. For each respective node out of said Ontology 1 node 11, 12, 13 and said Ontology 2 nodes 21, 22, 23 the network node 110 embeds the node description to capturing a semantic context associated with the node. This is to understand the similarity between the nodes to find the closeness of the node elements in the inter ontology perspective.

[0075] Action 205. The network node 110 further determines similarity scores between the embedded Ontology 1 node 11, 12, 13 descriptions and embedded Ontology 2 node 21, 22, 23 descriptions. A similarity score when used herein e.g. means represents the closeness of the nodes between two different ontologies. The closeness implies the mapping between two product nodes during migration.

[0076] According to the example, if migrating from any product to another product in a telecom domain and the nodes are referred to as fields, the network node 110 determines similarities between fields’ descriptions relating to the old product with the fields’ descriptions relating to the new product.

[0077] Action 206. The network node 110 establishes intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the Ontology 1 node 11 , 12, 13 descriptions and Ontology 2 node 21 , 22, 23 descriptions. An intra- ontology causality when used herein e.g. means a node element inside the node ontology which helps in defining the dynamics of the ontology node. These node elements causal relationships are established to understand the causal effects. The Intra ontology causal relationship describes the causal relationship between the nodes in an ontology.

[0078] In some embodiments, the establishing of the intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, is further based on the data associated with the respective nodes within each of Ontology 1 and Ontology 2. To discover the causal relationship between the nodes in an ontology data associated with the corresponding node may be used. With knowledge of the causal graphs for ontology 1 and ontology 2, these graphs may be used to find the map between these two ontologies.

[0079] The idea may be to understand the causal effect between intra ontologies and use this understanding to better mapping between 2 ontologies at node level.

[0080] Action 207. The network node 110 maps the ontologies. This is performed by mapping Ontology 1 nodes 11 , 12, 13 with Ontology 2 nodes 21 , 22, 23 based on the determined similarity scores and the established intra-ontology causality. This is e.g. to obtain matching ontology 1 and ontology 2 node pairs. The word map herein is in some embodiments e.g., used to understand the node descriptions similarity between two different ontologies.

[0081] The mapping of the ontologies by mapping Ontology 1 node 11 , 12, 13 with Ontology 2 nodes 21 , 22, 23 may in some embodiments further comprise the following actions 307a-c. The network node 110 may determine 307a a causality score for a node pair comprising an Ontology 1 node and an Ontology 2 node 21 , 22, 23. When the determined causality score is greater than or equal to a threshold, the network node 110 may map 307b the said Ontology 2 node 21 , 22, 23 as a match, e.g. as a best match, for said Ontology 1 node. When the determined causality score is lesser than the threshold, the network node 110 may discard 307c the said node pair. Action 208. The network node 110 may present the result of the mapping of the ontologies. This may be performed by showcasing interconnected ontologies nodes according to the similarity scores with semantic context and intra-ontology causality insights.

[0082] Action 209. In some example embodiments, the ontology mapping relates to a data migration relating to telecommunications operations within a telecom domain, e.g. covering migrating from a first product to a second product. In these embodiments, the first set of data may comprise data related to the first telecommunication product and the second set of data comprises data related to the second telecommunication product to be migrated to. In these embodiments, the network node 110 may further perform the migration from the first telecommunication product to the second telecommunication product based on the mapped ontologies. This is an advantage since having this type of system makes the product migration simpler and more effective.

[0083] Action 210. In some embodiments, the ontology mapping relates to a product migration in a telecom domain. This product migration may comprise subscriber migration from a first product related to handling of subscriber resource consumption, such as e.g. a legacy charging system, to a second product related to handling of subscriber resource consumption, such as e.g. a new charging system. In these embodiments, the first set of data may comprise data related to the first product, and the second set of data may comprise data related to the second product to be migrated to. In these embodiments, the network node 110 further performs the subscriber migration from the first product of handling telecom operations to the second product of handling telecom operations based on said mapped ontologies.

[0084] In some embodiments, the ontology mapping relates to mapping between a first healthcare dataset and a second healthcare dataset collected in different point of times. In these embodiments, the first set of data comprises data related to the first healthcare dataset, and the second set of data comprises data related to the second healthcare dataset. This will be described more in detail below.

[0085] Embodiments herein such as the embodiments mentioned above will now be further described and exemplified. The text below is applicable to and may be combined with any suitable embodiment described above. In this section, the provided architecture will be presented and discussed in a detailed manner. First the architecture and each of it’s individual components, also referred to as nodes, will be explored. This includes generating pairs among the ontologies, such as Ontology 1 and Ontology 2, with semantic similarity scores, finding out the causality inside each ontology separately and finally utilizing the semantic similarity scores between different pairs of the ontologies and the causality to find the best possible matches. Uses of implementing the methodology in the telecom sector, especially in the cases of product migration will be explained. Later the implementation of the provided method on two different datasets will be described, the second one more causally related than the first. Further various results of the methodology under varying scenarios will be presented. Finally, to conclude this section, some other use cases will be described where embodiments herein may be implemented.

[0086] As mentioned above, example embodiments herein provide methods that utilizes intra-ontology causality information inside an ontology to obtain an ontology mapping. A flowchart displaying the provided architecture is given in Figure 3. Figure 3 depicts a causality-powered ontology mapping architecture according to some embodiments herein, for a causal effect impacting product migration. The Causal relation helps in achieving accurate ontology mapping.

[0087] Semantic Similarity for Cross Ontology Mapping

[0088] One goal is to identify semantically similar nodes, also referred to as concepts and objects, between two different ontologies, such as Ontology 1 and Ontology 2. To achieve this, two different strategies may be used.

[0089] Utilizing Transformer Models: The descriptions of individual nodes within the ontologies play a pivotal role in understanding the semantics of these entities. To generate meaningful representations of these descriptions, the state-of-the-art transformer model, BERT may be used. By passing the node descriptions through BERT, embeddings that encapsulate the semantic context and information associated with each entity will be generated. Importantly, this process may extend beyond the mere text matching of entities and delves into the essence of their meaning. Generation of Descriptions: In cases where descriptions for ontology nodes are unavailable, large language models (LLMs) may be used, e.g., ChatGPT, Pathways Language model. These models enable automatic generation of descriptions by contextualizing the nodes and inferring their attributes and relationships, thus enhancing the comprehensiveness of the mapping process. This generation of description using LLMs is particularly novel in scenarios when the labels of the ontology classes contain very little information for transformer models to generate proper embeddings. In this It has been demonstrated above, the importance of generating the descriptions, field names have been considered to illustrate the effectiveness of embodiments herein.

[0090] A cosine similarity between the embedding vectors that were formed has been calculated, resulting in the creation of a similarity matrix. The most closely connected nodes across the two ontologies, such as Ontology 1 and Ontology 2, may be revealed by the highest matches in this matrix, providing a strong basis for cross-ontology mapping. With the assistance of a domain expert, a threshold for the cosine similarity value may be set to identify the best matches, allowing one to filter out all but the most suitable matches of nodes between the ontologies, such as Ontology 1 and Ontology 2. However, as discussed semantic based mapping itself is insufficient in certain cases to provide a suitable match, e.g., because of the false positives, there might be semantic similarity. But there is not any actual relation and false negatives, there is an actual relation but no match in semantic similarity. An intra ontology based causal discovery procedure may be used to enhance our semantic based matching accuracy.

[0091] Causality based Intra-Ontology Mapping

[0092] A causal discovery method may be used to discover the cause-and-effect relationships inside the given ontology, such as the Malinsky, Daniel, and David Danks. "Causal discovery algorithms: A practical guide." Philosophy Compass 13.1 (2018): e12470, referred to as the PC algorithm herein.

[0093] The PC algorithm works in the following way:

[0094] PC Algorithm:- Build an Independence Graph:

[0095] - For each pair of variables, test conditional independence is given to all possible subsets of other variables. The conditional independence equation is shown below, here random variables X and Y are said to be conditionally independent if they satisfy, they follow property.

[0096] P(X | Y,Z) = P(X\Z)

[0097] - Construct an undirected graph where nodes represent variables, and edges indicate conditional dependence.

[0098] Orient the Edges:

[0099] - Orient the edges in a way that is consistent with the conditional independence relations observed.

[0100] - This involves a set of rules to determine the orientation of edges.

[0101] Remove Spurious Edges:

[0102] - Eliminate edges that can be explained by conditioning on other variables.

[0103] Ontology Mapping Enhancing Semantic-Based Mapping Through Causal Neighbors

[0104] For a robust ontology mapping framework, an Ontology Mapping Technique designed to fortify the semantic-based mapping of nodes according to embodiments herein is introduced. Acknowledging that direct semantic similarities between nodes may encounter limitations, embodiments herein provide the integration of causal neighbours to offer a more nuanced and comprehensive mapping method. Parents and children of a node is its causal neighbour.

[0105] Causal Neighbours Integration: lhe provided method the network node 110 may consider both first-level parents and children to identify the mapping, where the understanding that causal parents and children of the matching nodes are mostly same and often share same underlying semantic contexts are considered.

[0106] Semantic Linkage Assessment:

[0107] For each node, the network node 110 examines the nodes causal neighbours and evaluates their semantic linkage to the corresponding neighbours of the target node (parent to parent and child to child). Leveraging the embedding vectors generated by transformer models, the network node 110 may utilize the semantic similarity scores between the neighbours of the two nodes. This dual assessment provides a quantifiable measure of how closely linked the causal neighbours are in terms of semantic content.

[0108] Reinforcing Mapping Confidence:

[0109] A reinforcement factor for mapping confidence may be derived by the network node 110 from the number of causal neighbours exhibiting semantic linkage, e.g. a score greater than the preset semantic threshold, to the neighbours of a target node. Essentially, the more causally connected neighbours that share similar semantic contexts, the more substantial the support for the validity and accuracy of the semantic-based mapping. This strategic integration serves as a powerful mechanism to strengthen the mapping results.

[0110] Addressing Semantic Ambiguities:

[0111] The Ontology Mapping Reinforcement Technique according to some embodiments herein addresses potential ambiguities arising from limited direct semantic overlap. In cases where traditional semantic-based mapping falls short, the introduction of causally linked neighbours serves as a valuable corrective measure, enhancing the overall reliability, precision, and comprehensiveness of the ontology mapping process.

[0112] A flowchart depicted in Figure 4 displays a flow of finding a mapping for the final ontology mapping according to some embodiments herein. As shown, first the semantic embeddings generated from the descriptions are evaluated on the basis of cosine similarity score. Then if the similarity score is greater than the threshold set by a user, the network node 110 proceeds to check causality score. In Figure 4 Ontology 1 is referred to as Ontology A and Ontology 2 is referred to as Ontology B. Figure 4 depicts node description as sentences which is used to map between two different nodes on 2 different ontologies. This may comprise that similarity score based on node description and Causality score based on intra ontology causal graph are exploited to obtain the accurate mapping.

[0113] Causality Score: To generate the causality score between a given pair of nodes, each from one of the ontologies, such as Ontology 1 and Ontology 2, the network node 110 may first identify the parents and children of both the nodes separately. When the network node 110 has obtained the parents and children of both the nodes the network node 110 may generate cross pairs, also referred to as Cross Product, between the parents of one node to the other and children of one node to the other. After having the pairs of parents and children, by utilizing the description based semantic embeddings generated earlier the network node 110 may get the cosine similarity scores between each of these pairs. The network node 110 may then use the preset threshold to find the number of parents and children that are semantically similar to their counterparts. The total number of parents and children that have a semantically similar counterpart is the causality score between that particular pair of nodes.

[0114] The network node 110 may use two separate causality thresholds alpha and beta, alpha causality thresholding is applied to the mappings that pass the semantic threshold filtering and the beta score is applied for the mappings that do not. The alpha score may usually be lesser than the beta thresholding score to allow mappings that pass the semantic test into the final mapping. The beta score may be set at a higher value to strictly filter out the matches that do not pass the semantic threshold, to allow the mappings that our semantic mapping system wrongly classifies and are true matches between the ontologies.

[0115] Finally, those matches that pass the causality and semantic thresholds are selected to be the part of our final ontology mapping.

[0116] It should be noted that the thresholds used in the architecture provided by embodiments herein may be selected empirically in different situations with the help of domain knowledge. A proper selection of these thresholds holds the key to efficient working of this architecture.

[0117] An example of a pseudo-code of the final algorithm comprises the following:

[0118] In the text herein Ontology 1 may be referred to as Ontology A and Ontology 2 may be referred to as Ontology B.

[0119] # Step 1 : Generate Cross Pairs cross_pairs = generate_cross_pairs (ontology_A, ontology_B)

[0120] # Step 2: Initialize Final Mapping Set final_mapping = set() # Step 3: Iterate through Cross Pairs for pair in cross_pairs:

[0121] {

[0122] # Step 4: Evaluate Semantic Similarity semantic_similarity = calculate_cosine_similarity(pair)

[0123] # Step 5: Check Semantic Threshold semantic_threshold = user_set_semantic_threshold() if semantic_similarity >= semantic_threshold:

[0124] {

[0125] # Step 6: Calculate Causality Score causality_score = calculate_causality_score(pair)

[0126] # Step 7: Check Alpha Threshold alpha_threshold = user_set_alpha_threshold() if causal ity_score >= alpha_threshold:

[0127] # Step 8: Apply Alpha Threshold

[0128] {

[0129] # Step 9: Add to Final Mapping final_mapping.add(pair)

[0130] }

[0131] } else:

[0132] {

[0133] # Step 6: Calculate Causality Score causality_score = calculate_causality_score(pair)

[0134] # Step 7: Check Beta Threshold beta_threshold = user_set_beta_threshold() if causal ity_score >= alpha_threshold:

[0135] # Step 8: Apply Beta Threshold

[0136] {

[0137] # Step 9: Add to Final Mapping final_mapping.add(pair)

[0138] } }

[0139] }

[0140] Integration of Mapping and Causality

[0141] Once completed both inter-ontology mapping with transformer models and intraontology causality mapping and enhancing the semantic based linkages with causality score, a last stage may entail integrating these essential elements. The network node 110 may integrate the information of the Mapping enhanced by causality and the causal discovery by presenting our results in an interactive graph-based structure which displays the different inter and intra ontology linkages between the different concepts of ontologies. This way of representation might not be feasible in larger ontologies, however in relatively smaller ontologies where the linkages may be clearly studied. This method will lead to Actionable insights which can better the decision-making process. Overall, this methodology holds the potential to close knowledge gaps inside each individual ontology as well as structural gaps between ontologies, revolutionizing data integration, knowledge representation, and decision-making processes in ontologies of any size. Further, information about the methodology, data sets used, outcomes, and ramifications of this approach may be provided in a Business Support System (BSS) product migration sections that follows.

[0142] BSS - Product migration:

[0143] One of the applications that this embodiment will have is in the problem of product migration in the telecom domain. In the complex landscape of product migration, a smooth transition without operational disruptions is important. This is where mapping and comprehending causal relationships within a single ontology become valuable. Product migration often involves disparate systems with distinct ontologies to represent components, configurations, and dependencies. Cross-ontology mapping standardizes terminology and definitions, ensuring a shared understanding of components and their interrelationships. This standardization is essential for a successful migration. Additionally, within the migrated product environment, understanding causality is critical. Intra-ontology causality discovery reveals causal relationships between components, facilitating a clear understanding of how changes to one element may impact others. This knowledge empowers better risk assessment and enables the development of strategies to mitigate potential disruptions during the migration process. For example, consider a scenario of subscriber migration from one system relating to Ontology 1 of charging to another system relating to Ontology . Hese, the integration of a Causality-Powered Ontology Architecture provides a comprehensive solution to the challenges faced during product migration in the telecom domain, some of the key situations where it helps is.

[0144] Inter-System Mapping: The architecture according to some embodiments herein enables the mapping of subscribers and their service plans between legacy and new systems. Semantic-based ontology mapping using Transformer models ensures a high degree of accuracy in this process.

[0145] Causal Analysis: Causal discovery within individual ontologies may allow operators to understand the causal relationships between service plan changes and subscriber experiences. This insight may be crucial for predicting and mitigating potential disruptions.

[0146] Churn Reduction: By applying causal discovery techniques, operators may identify the causes of subscriber churn during migration. This understanding allows for proactive measures to retain subscribers. For example, operators can offer targeted incentives or support to subscribers at high risk of churning.

[0147] Resource Allocation Mapping: The architecture according to some embodiments herein may support the mapping of network resources and billing systems between legacy and new platforms. By combining ontology mapping and causal analysis, operators may ensure efficient resource allocation based on causal insights.

[0148] Causality-Powered Insights: The architecture according to some embodiments herein may provide operators with a deep understanding of the impact of product migration on various aspects of telecom operations. This knowledge allows for predictive analytics to anticipate potential issues and plan for them in advance.

[0149] Data-lnformed Decision-Making: The integration of ontology mapping with causality analysis according to some embodiments herein may empower decision-makers with a holistic view of migrating subscriber from one product to another. It may enable informed decisions based on both structural mappings and causal relationships.

[0150] Figure 5 depicts an example of Mapping Using Proposed Architecture on Telecom Ontologies. The nodes of Ontology 1 are referred to as Plan A, Plan B and Plan C. The nodes of Ontology 2 are referred to as Subscription 1, Subscription 2 and Subscription 3. Figure 6 depicts parts of a semantic based mapping on existing descriptions on charging and beam objects with a threshold value of 0.65. Figure 6 shows the use for node level comparisons between ontologies.

[0151] It is observed that causality reinforces ontology mapping and gives a better mapping for product migration process. To explain the concept in a simple and effective manner, an easily understandable domain “Healthcare dataset is taken” to demonstrate an implementation according to some embodiments herein. The same may be easily implementable towards product migration task.

[0152] Implementation View

[0153] The provided example method is demonstrated by considering a task of mapping between two healthcare datasets collected in different years. The uniqueness of this dataset pair lies in the evolution of data collection, wherein the latter dataset has expanded in size due to the inclusion of additional variables for comparison. The datasets under consideration encompass crucial healthcare indicators and are categorized into two distinct ontologies, Ontology A and Ontology B. Ontology A represents data from the dataset of Annual Health Survey in the year 2010-2011 and Ontology B Represents a comparison dataset of the health survey features between 2010-2011 and 2011-2012 containing some of the data from the first dataset also addition data for future year as well as an additional comparison column called change in points between the two years. The fields available in both the ontologies are given below.

[0154] Ontology A :- ['Sr. No.', 'State', 'District', 'CBR', 'CDR', 'IMR', 'Neo- natal Mortality Rate', 'Under Five Mortality Rate ', 'MMR', 'Sex Ratio at Birth (SRB)', 'Sex Ratio (0-4 Years)', 'Sex Ratio (Total)’]

[0155] Ontology B:-['State', 'District', 'Crude Birth Rate (CBR)-2010-1 T, 'Crude Birth Rate (CBR)-2011-12', 'Crude Birth Rate (CBR)-Change in points', 'Crude Death Rate (CDR)- 2010-1 T, 'Crude Death Rate (CDR)-2011-12', 'Crude Death Rate (CDR)-Change in points', 'Infant Mortality rate (IMR)-2010-1 T, 'Infant Mortality rate (IMR)-2011-12', 'Infant Mortality rate (IMR)-Change in points', 'Neo- natal Mortality Rate-2010-1 T, 'Neo- natal Mortality Rate-2011-12', 'Neo- natal Mortality Rate-Change in points', 'Under Five Mortality Rate-2010-1 T, 'Under Five Mortality Rate-2011-12', 'Under Five Mortality Rate- Change in points', 'Maternal Mortality Rate (MMR)-2010-1 T, 'Maternal Mortality Rate (MMR)-2011-12', 'Maternal Mortality Rate (MMR)-Change in points', 'Sex Ratio at Birth (SRB)-2010-1 T, 'Sex Ratio at Birth (SRB)-2011-12', 'Sex Ratio at Birth (SRB)-Change in points', 'Sex Ratio (0-4 Years)-2010-1 T, 'Sex Ratio (0-4 Years)-2011-12', 'Sex Ratio (0-4 Years)-Change in points', 'Sex Ratio (Total)-2010-1 T, 'Sex Ratio (Total)-2011-12', 'Sex Ratio (Total)-Change in points']

[0156] Ontology Formation and Description Generation

[0157] The first step in our quest is to create two alternative ontologies, one for each dataset. As the nodes in our ontological structures, we take the column names extracted from both datasets to enable meaningful mapping. Nonetheless, we use ChatGPT, a cutting-edge language modelling tool, to improve the depth of our mapping procedure. Rich semantic context for mapping is produced by ChatGPT, which creates thorough descriptions for every field in the ontologies.

[0158] For example, the descriptions generated by ChatGPT for Ontology A is as follows

[0159] Ontology: - ['Sr. No.', 'State', 'District', 'CBR', 'CDR', 'IMR', 'Neo- natal Mortality Rate', 'Under Five Mortality Rate ', 'MMR', 'Sex Ratio at Birth (SRB)', 'Sex Ratio (0-4 Years)', 'Sex Ratio (Total)’]

[0160] Descriptions: - [

[0161] "Serial Number or Identifier",

[0162] "The geographical region within a country or nation",

[0163] "A defined administrative region within a state or country",

[0164] "The number of live births per 1 ,000 people in a given population in a specific time period",

[0165] "The number of deaths per 1 ,000 people in a given population in a specific time period",

[0166] "The number of infant deaths (under one year of age) per 1 ,000 live births in a specific time period",

[0167] "The number of deaths of infants within the first 28 days of life per 1 ,000 live births in a specific time period",

[0168] "The number of deaths of children under the age of five per 1 ,000 live births in a specific time period",

[0169] "The number of maternal deaths related to pregnancy or childbirth per 100,000 live births in a specific time period",

[0170] "The ratio of male to female births in a given population, often expressed as the number of male births per 1 ,000 female births", "The ratio of males to females among children aged 0 to 4 years in a given population",

[0171] "The overall ratio of males to females in a given population"

[0172] Sentence Embeddings and Similarity Score

[0173] After obtaining descriptions, a Sentence BERT is used as a transformer model. Sentence BERT Roberta Large creates sentence embeddings for every description sentence by utilizing Siamese BERT networks in an extremely efficient implementation. The semantic core of any description is captured by these embeddings. Using the cosine similarity measure, we compute similarity scores across these embeddings, enabling us to determine the top three matches for each node in the ontologies.

[0174] Finally, some embodiments herein also utilize a similarity score threshold of 0.7 to filter the matches. This Similarity score can be decided by looking at all of the matches with the help of a domain expert and deciding the similarity level above which the mapping between the node descriptions is relevant. This process may ensure that only the most relevant nodes are considered for mapping.

[0175] Causal Discovery Using PC Algorithm

[0176] At the same time that the inter-ontology mapping procedure is developing causal linkages may be identified within each individual ontology, such as Ontology 1 and Ontology 2. Here, the network node 110 may make use of the PC algorithm, a reputable and popular causal discovery method. By using graph theory and conditional independence tests. The PC method may create a causal graph with directed edges that graphically depicts causal links. It is quite good at identifying causal relationships, which improves the comprehension of the causation that exists in every ontology. The final conclusions may be enhanced by this integrated method, which offers a comprehensive perspective of data linkages and their underlying causes.

[0177] Integration of Intra and Inter Ontology Mapping

[0178] The methodology may culminate in a smooth amalgamation of the intra-ontology causality mapping and the transformer model-based inter-ontology mapping. A strong and thorough final mapping is produced by combining intra- and inter-ontology insights, providing a distinctive viewpoint on data relationships. Implementation: Synthesizing and Mapping Causally Related Datasets to demonstrate causal reinforcement

[0179] In order to validate and showcase the effectiveness of the Causality reinforce Ontology Architecture, another implementation phase according to some embodiments herein is demonstrated, involving a causally related dataset this time, specifically focusing on a Lung Cancer dataset. The objective is to synthesize another dataset by duplicating the original and introducing variations. This synthesized dataset will serve as a testbed for our Description-Generated Semantic-Based Mapping and the introduced Causal Reinforcement Technique.

[0180] Dataset Synthesis and Modification:

[0181] Duplication: The network node 110 may begin by duplicating the Lung Cancer dataset, effectively creating a copy that retains the causally related structure of the original.

[0182] Introduction of Variations: To simulate an evolving dataset, random variations are introduced by adding a few columns and altering some column names. These changes mimic the natural progression of data collection, as observed in real-world scenarios.

[0183] Ontology Formation and Description Generation:

[0184] Utilizing the duplicated and modified Lung Cancer dataset, the ontology formation process outlined in our architecture is followed. The column names serve as nodes in the ontological structures. To enrich the mapping procedure, ChatGPT is leveraged for generating detailed descriptions for each field, ensuring a semantic context for mapping.

[0185] Description-Generated Semantic-Based Mapping:

[0186] Sentence Embeddings: Sentence BERT is employed to create sentence embeddings for the descriptions generated by ChatGPT. This step captures the semantic core of each field.

[0187] Similarity Score Calculation: Using cosine similarity measures, similarity scores are calculated across the embeddings. This enables to identify the top matches for each node in the ontologies.

[0188] Causal Reinforcement Technique:

[0189] In this unique implementation, the Causal Reinforcement Technique according to some embodiments herein may be integrated to further enhance the mapping accuracy: Dataset Duplication: The causally related nature of the Lung Cancer dataset allows to create causal links between corresponding nodes in both the original and duplicated datasets.

[0190] Causal Neighbour Identification: The causal neighbours within each dataset are identified, exploring first-level parents and children.

[0191] Semantic Linkage Assessment: Leveraging the ontology descriptions and embeddings, the semantic linkage between causally connected neighbours (parents and children) is assessed. The mapping is repaired and reinforced by creating the mappings even though sometimes the semantic mappings fail.

[0192] Reinforcing Mapping Confidence: The number of causally connected neighbours exhibiting strong semantic linkage reinforces the mapping confidence, that may ensure a more resilient and accurate ontology mapping.

[0193] Benefits of Synthesis and Mapping:

[0194] This implementation according to some embodiments herein not only validates the adaptability of the architecture to causally related datasets but also demonstrates its efficacy in handling variations and synthesizing meaningful mappings. The results section will provide insights into the architecture's robustness and its applicability in scenarios where ontologies evolve over time and the node names change, showcasing the potential impact on decision-making processes in data analytics.

[0195] In the subsequent sections, the empirical results will be presented, the reinforced mappings will be analysed, and the broader implications of the approach in the context of different datasets will be discussed.

[0196] The method according to some embodiments herein, which combines causal discovery with ontology mapping, has produced thorough and reliable results that shed light on the connections between and within the selected healthcare datasets. The results of applying the architecture some embodiments herein without the causal reinforcement technique in the original and on some other different scenarios are shown further in this document. The demonstration of the causal reinforcement technique is shown on a specialized causality dataset after a discussion related to the results from the healthcare statistics dataset. Inter-Ontology Mapping (Short Dashed Line Undirected Edges): The results of the inter-ontology mapping process are shown by the Short Dashed Line undirected edges in the visualization. Beyond structural similarities, these lines provide a thorough semanticbased mapping by defining the links and connections between nodes in two different ontologies.

[0197] Intra-Ontology Mapping (Thin Solid Lines and Long Dashed Lines): The thin solid and long dashed lines serve to enhance the findings of the inter-ontology mapping. The causal linkages found within Ontology A by using the PC technique are represented by the Thin Solid lines and the ones for Ontology B are represented using the Long Dashed lines. Notably, a "G2_ " and Circular Nodes + prefix makes it easy to identify nodes from Ontology B, which improves the clarity of the findings.

[0198] Based on the data, it may be concluded that the semantic based ontology mapping (Short Dashed Lines) is yielding good results in terms of mapping between the ontologies. The results also show that the intra ontology causal discovery does rather well in determining the causal linkages inside the dataset.

[0199] A comprehensive understanding of data links is provided by the combination of intra- and inter-ontology mapping, which unifies structural mapping with causal insights. This combination produces thorough and reliable mapping findings, which are essential for making well-informed decisions in the dynamic field of data analysis.

[0200] Ontology Mapping Efficiency and Robustness

[0201] To demonstrate the efficiency and robustness of our ontology mapping system we generated various results using two different scenarios.

[0202] 1 . Scenario 1 is where no descriptions are generated for the fields of the ontology and the architecture is just applied by generating semantic embeddings on the field names alone

[0203] 2. In Scenario 2 alternate field names are utilized that are semantically similar but lexically different to check how the architecture performs in ontology mapping under various scenarios, the field names ontology 2 are changed to the following names :- ['State', 'District', 'Natality Rate(NR)-2010-1 T, 'Natality Rate(NR)-2011-12', 'Natality Rate(NR)-Change in points', 'Mortality Rate (MR)-2010-1 T, 'Mortality Rate (MR)-2011-12', 'Mortality Rate (MR)-Change in points', 'Infant Death rate-2010-1 T, 'Infant Death rate- 2011-12', 'Infant Death rate-Change in points', 'Neo- natal Death Rate-2010-1 T, 'Neonatal Death Rate-2011-12', 'Neo- natal Death Rate-Change in points', 'child under-five mortality rate-2010-1 T, 'child under-five mortality rate-2011-12', 'child under-five mortality rate-Change in points', 'Maternal death rate-2010-1 T, 'Maternal death rate-2011-12', 'Maternal death rate-Change in points', 'Birth gender ratio-2010-1 T, 'Birth gender ratio- 2011-12', 'Birth gender ratio-Change in points', 'child sex ratio (0-4 years)-2010-1 T, 'child sex ratio (0-4 years)-2011-12', 'child sex ratio (0-4 years)-Change in points', 'overall sex ratio-2010-1 T, 'overall sex ratio-2011-12', 'overall sex ratio-Change in points']

[0204] Summary of the results

[0205] Table 2

[0206] The result is presented in Table 2 above. The results use two metrics precision and recall which make use of 3 parameters True Positives, False Negatives and False Positives, the formulae for precision and recall are as follows

[0207] True Positives

[0208] Precision = - - -

[0209] True Positives + raise Positives

[0210] True Positives

[0211] Recall = — - - — — - — - - - ■_ -

[0212] True Positives + False Negatives

[0213] Results Inference: Impact of Description Generation on Ontology Mapping Accuracy.

[0214] The obtained precision and recall metrics for the ontology mapping results, presented in Table 2 above, underscore the significance of incorporating field descriptions generated by ChatGPT in the causality-powered ontology architecture. The evaluation compares scenarios with and without descriptions, considering both the original and alternate names for columns.

[0215] Effect of Description Generation: The inclusion of descriptions significantly improves both precision and recall metrics in ontology mapping. With descriptions, precision increases from 0.64 to 0.81 , and recall improves from 0.55 to 0.9, showcasing the vital role played by detailed field descriptions in enhancing mapping accuracy.

[0216] Robustness to Name Changes: Even in scenarios where column names are altered (Alternate Names), the architecture demonstrates remarkable resilience. The precision remains consistently high at 0.8, while recall reaches a perfect score of 1. This indicates that the ontology mapping process is not solely reliant on exact column name matches, proving the adaptability and effectiveness of our approach.

[0217] Inference:

[0218] The noteworthy improvement in precision and recall with the inclusion of descriptions highlights the crucial role played by Generative Al in generating rich semantic context for ontology nodes. This context allows the architecture according to some embodiments herein to transcend the limitations imposed by changes in column names. The architecture's ability to map accurately even when confronted with alternate names underscores its robustness and adaptability.

[0219] These findings emphasize the importance of investing in comprehensive description generation as part of ontology mapping processes. By capturing the semantic essence of each field, our architecture not only overcomes challenges posed by name changes but also enhances the overall precision and recall, making it a valuable tool for data integration and decision-making in dynamic and evolving datasets.

[0220] With the use of some examples from the findings, it will now be discussed how the architecture according to some embodiments herein is very helpful in improving the ontology alignment methods that are currently in use.

[0221] Causality Discussion

[0222] This section will showcase a selection of images that represent subsets of the outcomes obtained from utilising the architecture according to some embodiments herein on the previously stated datasets. It will be illustrated in the following Figures how ontology mapping is enhanced by the idea of intra-ontology causal discovery.

[0223] Figure 7 depicts example 1 according to some embodiments herein. A simple causal link between infant mortality rate change in points and under five mortality rate change in points has been described above. It is known that this link is correctly detected for a fact because it is known that infant mortality influences under five mortality rate and not the other way around and our causal discovery algorithm discerns the causality accurately in this situation. This example is required to demonstrate how embodiments of the architecture successfully discovers causality and helps in making the ontology mapping more detailed. Figure 8 depicts example 2 according to some embodiments herein.

[0224] With help of the semantic-based alignment technique, the sex ratios from the two separate years are mapped to the first ontology's sex ratio and described above. But the algorithm used for causal discovery also found that there was a causal relationship between the two years' worth of data. This example suggests that mapping linkages inside an ontology that were previously unknown may be found by combining causality with ontology mapping.

[0225] Figure 9 depicts example 3 according to some embodiments herein.

[0226] The Crude Death Rate (CDR) and the Crude Birth Rate (CBR) are mapped between two ontologies in this example as described above. Although, the ontology mapping system was unable to find two links to the years of CDR and CBR, the link is implicitly inferred because causal discovery was also used. In this instance, even if one of the embodiments of the ontology mapping systems is unable to identify certain linkages, the causal discovery used in conjunction with ontology mapping provides fault tolerance to an extent by identifying the relevant mapping by itself.

[0227] By identifying causal relationships in large systems where a change in the system variable may have cascading effects and enhancing the decision-making process, the results and discussion section has hopefully been sufficiently demonstrated how the provided architecture improves upon the existing ontology mapping systems. Embodiments of the system may be scaled to offer mapping and causation between big ontologies, even though real ontology mapping systems operate with far larger datasets than the datasets we have utilized.

[0228] Causal Reinforcement on LUCAS dataset Results

[0229] In a practical demonstration of the effectiveness of some embodiments of the architecture, it was applied to the Land Use / Cover Area frame statistical Survey (LUCAS) dataset, a dataset known for its intricate causal relationships. In addition to standard semantic-based mapping, an embodiments of the Causal Reinforcement Procedure were implemented to strengthen the links between ontologies using intra-ontology causal mapping.

[0230] Steps in Causal Reinforcement: Dataset Synthesis Application: The Lucas dataset was Utilized as the experimental dataset, known for its rich causal structure. The LUCAS dataset was synthesized by duplicating the dataset and changing the column names and adding a few random columns to further demonstrate the accuracy of our proposed methodology.

[0231] Description Generation and Semantic Mapping: The ontology mapping framework was applied for semantic links between nodes from different ontologies after generating descriptions for all the required nodes using generative Al.

[0232] Intra-Ontology Causal Mapping: The PC algorithm was leveraged to identify causal relationships within each individual ontology.

[0233] Causal Reinforcement Procedure: For each node, the causal neighbours (parents and children) was matched with every other node, establishing causal links between nodes.

[0234] Integration of Semantic and Causal Mapping: The inter-ontology mapping was based on both semantic and causal mappings, ensuring a comprehensive and reinforced result.

[0235] Result Presentation: The mapping results were displayed with an innovative visualization approach. Thicker links between nodes indicated the reinforcement of the link by causal neighbours. This visual representation showcased the degree to which each link was strengthened by the causal neighbours from one side being mapped to causal neighbours from the other side.

[0236] Benefits and Insights:

[0237] Enhanced Link Reliability: The integration of intra-ontology causal mapping according to embodiments herein reinforced the semantic links, enhancing the overall reliability and accuracy of the ontology mapping.

[0238] Visual Reinforcement: The visual presentation with thicker links provided an intuitive understanding of the reinforced links, offering insights into the interconnected nature of data relationships.

[0239] Adaptability to Complex Structures: The successful application of embodiments of the Causal Reinforcement Procedure to the Lucas dataset demonstrated the adaptability of our architecture to datasets with intricate causal structures.

[0240] This real-world application of some embodiments of the Causality-Powered Ontology Architecture in the context of the Lucas dataset exemplifies its ability to navigate complex causal relationships and provide nuanced insights into the mapping of ontologies. The visual reinforcement approach offers a transparent depiction of the enhanced links, emphasizing the robustness and practicality of our innovative methodology.

[0241] Figure 10 discloses an example embodiment of the Architecture Applied on LUCAS dataset with causal reinforcement. A final mapping after applying the ontology mapping methodology is shown in Figure 10. The circular nodes are from Ontology A and the circular nodes having G2 in their label name, are from ontology B. Intra ontology causality in ontologies A and B are displayed by thin solid lines and short dashed links respectively. The inter ontology mapping is shown by the short, dashed lines which are of varying thicknesses, the thicker links being more causally reinforced by more causal neighbours of the nodes being semantically mapped.

[0242] In the figure, there is a also a thicker solid edge between Car_accident and road_crash. The existence of this edge in this situation is an indication of a Mapping defined by the causal neighbours of the nodes being semantically related with each other. This mapping is highlighted separately because in this situation where our description based semantic mapping fails to identify the linkage, there is a contingency measure in the form of the causal reinforcement techniques that gives an accurate mapping.

[0243] Use case:

[0244] Vehicular Computing (ITS scenario)

[0245] The development of Intelligent Transportation Systems (ITSs) faces the challenge of integrating data from multiple unrelated sources. As one of the core technologies of knowledge integration in ITS, an ontology typically provides a normative definition of transportation domain that may be used as a reference for information integration. However, due to the subjectivity of domain experts, a concept may be expressed in multiple ways, yielding the ontology heterogeneity problem. Ontology matching (OM) is an effective method of addressing it, which is of help to further realize the mutual communication between the ontology based ITS. In order to achieve mutual communication between ontology based ITSs, it is important to determine the logical relationships between heterogeneous ontologies. OM is an important technique to solve the problem of semantic heterogeneity, which is dedicated to discovering correspondences between related entities (e.g., classes and properties) in different ontologies. For this reason, it is effective to use ontology matching techniques to solve the existing semantic heterogeneity of transport ontology. The transmodel is a reference data model about the European public transport information system that represents traffic ontologies and their relationships. Here, to establish effective ontology matching, embodiments herein will yield better results towards knowledge integration process in ITS.

[0246] Ticket Handling Process

[0247] Additionally an experiment of ticket handling perform within a Charging node. Here, the Idea is to express the knowledge of how the similar ticket are handled by different nodes. This idea studies how to express these steps of solving the ticket as ontology by establishing similar domain mapping. This idea validates the benefit of using ontology in multimode scenarios especially when data is distributed. The above entity relationship can also be expressed as a knowledge graph. It is also possible to verify cause and effect between the Knowledge Graph so that we can be able to express it as causal knowledge graph. Now the specific scenario where some embodiments of the methodology is tried out may be considered.

[0248] Logical Volume Module (LVM) setup has been considered to which is handled in some nodes. Below are the steps followed by an Online Charging System (OCS) node to fix the LVM setup.

[0249] Create a logical volume group.

[0250] Create a physical volume group.

[0251] Update the above groups in the system configuration.

[0252] Run the cfengine to update specified and actual configuration. A cfengine when used herein may e.g. be a configuration engine in the framework architecture.

[0253] The above steps may be well explained in the form of multiple entities. It may also be expressed as ontology by creating a relationship between the entities.

[0254] The above method is only applicable to an OCC node but not to other nodes like SDP, CRS and NGCS etc. The reason is because the entity setup for each node is independent of the other.

[0255] As mentioned above, each node has a different entity setup, and each carries its own ontology structure. It is tried to map the different ontologies which may give a generic and best solution for a new ticket. Understanding causality on each ontology will be helpful to connect the ontologies with good number of positive links which establishes effective mapping and provide required knowledge to solve the ticket. The network node 110 and all methods provided herein may be implemented as cloud applications, running in virtualized environments, or in distributed servers.

[0256] To perform the method actions above, the network node 110 is configured to map ontologies in a communications network 100.

[0257] The network node 110 may comprise an arrangement depicted in Figure 11. The network node 110 may comprise an input and output interface 1100 configured to communicate in the communications network 100. The input and output interface 1100 may comprise a wireless receiver not shown, and a wireless transmitter not shown.

[0258] The network node 110 is further being configured to: Obtain a first ontology, Ontology 1 , adapted to represent data of a first set of data, and a second ontology, Ontology 2, adapted to represent data of a second set of data.

[0259] The network node 110 is further being configured to:- Collect a number of Ontology 1 node 11, 12, 13 from the data of the first set of data, wherein each respective Ontology 1 node comprises data from the first set of data, and a number of Ontology 2 nodes 21, 22, 23 from the data of the second set of data, wherein each respective Ontology 2 nodes 21 , 22, 23 comprises data from the second set of data.

[0260] The network node 110 is further being configured to, for each respective node out of said Ontology 1 node 11, 12, 13 and said Ontology 2 nodes 21, 22, 23,

[0261] - Generate a node description based on a content of the data in the node, and

[0262] - Embed the node description, capturing a semantic context associated with the node.

[0263] The network node 110 is further being configured to,

[0264] - Determine similarity scores between the embedded Ontology 1 node descriptions and embedded Ontology 2 node 21, 22, 23 descriptions,

[0265] - Establish intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the Ontology 1 node descriptions and Ontology 2 node 21 , 22, 23 descriptions, and

[0266] - Map the ontologies by mapping Ontology 1 node 11 , 12, 13 with Ontology 2 nodes 21 , 22, 23 based on the determined similarity scores and the established intra-ontology causality.

[0267] In some embodiments, the ontology mapping is adapted to relate to a data migration relating to telecommunications operations within a telecom domain. In these embodiments, the first set of data is adapted to comprise data related to a first telecommunication product, and the second set of data adapted to comprise data related to the second telecommunication product to be migrated to. In these embodiments, the network node 110 is further configured to perform the migration from the first telecommunication product to the second telecommunication product based on the mapped ontologies.

[0268] In some embodiments, the ontology mapping is adapted to relate to a product migration in a telecom domain, comprising subscriber migration from a first product related to handling of subscriber resource consumption, to a second product related to handling of subscriber resource consumption. In these embodiments, the first set of data is adapted to comprise data related to the first product, and the second set of data adapted to comprise data related to the second product to be migrated to. In these embodiments, the network node 110 is further configured to perform the subscriber migration from the first product of handling telecom operations to the second product of handling telecom operations based on said mapped ontologies.

[0269] In some embodiments, the ontology mapping is adapted to relate to mapping between a first healthcare dataset and a second healthcare dataset collected in different point of times. In these embodiments, the first set of data is adapted to comprise data related to the first healthcare dataset, and the second set of data adapted to comprise data related to the second healthcare dataset.

[0270] In some embodiments, the network node 110 is further configured to map the ontologies by mapping Ontology 1 node 11, 12, 13 with Ontology 2 nodes 21 , 22, 23 is by:

[0271] - Determine a causality score for a node pair comprising an Ontology 1 node and an Ontology 2 node 21, 22, 23,

[0272] - When the determined causality score is greater than or equal to a threshold, map the said Ontology 2 node 21, 22, 23 as a match, for said Ontology 1 node, and

[0273] - When the determined causality score is lesser than the threshold, discard the said node pair.

[0274] In some embodiments, the network node 110 is further configured to establish the intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the data associated with the respective nodes within each of Ontology 1 and Ontology 2.

[0275] In some embodiments, the network node 110 is further configured to present the result of the mapping of the ontologies, by showcasing interconnected ontologies nodes according to the similarity scores with semantic context and intra-ontology causality insights. Embodiments herein may be implemented through a respective processor or one or more processors, such as the respective processor 1110 of a processing circuitry in the network node 110 depicted in Figure 11 together with respective computer program code for performing the functions and actions of the embodiments herein. The program code mentioned above may also be provided as a computer program product, for instance in the form of a data carrier carrying computer program code for performing the embodiments herein when being loaded into the respective network node 110. One such carrier may be in the form of a CD ROM disc. It is however feasible with other data carriers such as a memory stick. The computer program code may furthermore be provided as pure program code on a server and downloaded to the respective network node 110.

[0276] The network node 110 may further comprise a respective memory 1120 comprising one or more memory units. The respective memory 1120 comprises instructions executable by the processor in the respective network node 110. The respective memory 1120 is arranged to be used to store e.g., media functions, indications, tags, information, data, configurations, communication data, and applications to perform the methods herein when being executed in the respective network node 110.

[0277] In some embodiments, a respective computer program 1130 comprises instructions, which when executed by the respective at least one processor 1110, cause the at least one processor of respective network node 110 to perform the actions above.

[0278] In some embodiments, a respective carrier 1140 comprises the respective computer program 1130, wherein the respective carrier 1140 is one of an electronic signal, an optical signal, an electromagnetic signal, a magnetic signal, an electric signal, a radio signal, a microwave signal, or a computer-readable storage medium.

[0279] Those skilled in the art will appreciate that units in the respective network node 110 described above may refer to a combination of analog and digital circuits, and / or one or more processors configured with software and / or firmware, e.g. stored in the respective network node 110, that when executed by the respective one or more processors such as the processors described above. One or more of these processors, as well as the other digital hardware, may be included in a single Application-Specific Integrated Circuitry ASIC, or several processors and various digital hardware may be distributed among several separate components, whether individually packaged or assembled into a System-on-a-Chip (SoC). When using the word "comprise" or “comprising” it shall be interpreted as nonlimiting, i.e. meaning "consist at least of'. The embodiments herein are not limited to the preferred embodiments described above. Various alternatives, modifications and equivalents may be used.

Claims

CLAIMS1 . A method performed by a network node (110), for mapping ontologies in a communications network (100), the method comprising: obtaining (201) a first ontology, Ontology 1 , representing data of a first set of data, and a second ontology, Ontology 2, representing data of a second set of data, collecting (202) a number of Ontology 1 nodes (11 , 12, 13) from the data of the first set of data, wherein each respective Ontology 1 node (11 , 12, 13) comprises data from the first set of data, and a number of Ontology 2 nodes (21 , 22, 23) from the data of the second set of data, wherein each respective Ontology 2 node (21 , 22, 23) comprises data from the second set of data, for each respective node out of said Ontology 1 nodes (11 , 12, 13) and said Ontology 2 nodes (21 , 22, 23): - generating (203) a node description based on a content of the data in the node, and - embedding (204) the node description, capturing a semantic context associated with the node, determining (205) similarity scores between the embedded Ontology 1 node (11 , 12, 13) descriptions and embedded Ontology 2 node (21 , 22, 23) descriptions, establishing (206) intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the Ontology 1 node (11 , 12, 13) descriptions and Ontology 2 node (21 , 22, 23) descriptions, and mapping (207) the ontologies by mapping Ontology 1 nodes (11 , 12, 13) with Ontology 2 nodes (21 , 22, 23) based on the determined similarity scores and the established intra-ontology causality.

2. The method according to claim 1 , wherein: the ontology mapping relates to a data migration relating to telecommunications operations within a telecom domain, the first set of data comprises data related to a first telecommunication product, and the second set of data comprises data related to the second telecommunication product to be migrated to, performing (209) the migration from the first telecommunication product to the second telecommunication product based on the mapped (207) ontologies.

3. The method according to any of the claims 1-2, wherein:the ontology mapping relates to a product migration in a telecom domain, comprising subscriber migration from a first product related to handling of subscriber resource consumption to a second product related to handling of subscriber resource consumption, the first set of data comprises data related to the first product, and the second set of data comprises data related to the second product to be migrated to, and performing (210) the subscriber migration from the first product of handling telecom operations to the second product of handling telecom operations based on said mapped (207) ontologies.

4. The method according to claim 1 , wherein: the ontology mapping (207) relates to mapping between a first healthcare dataset and a second healthcare dataset collected in different point of times, the first set of data comprises data related to the first healthcare dataset, and the second set of data comprises data related to the second healthcare dataset.

5. The method according to any of the claims 1-4, wherein the mapping (207) of the ontologies by mapping Ontology 1 nodes (11 , 12, 13) with Ontology 2 nodes (21, 22, 23) further comprises: determining (207a) a causality score for a node pair comprising an Ontology 1 node (11 , 12, 13) and an Ontology 2 node (21, 22, 23), when the determined causality score is greater than or equal to a threshold, mapping (207b) the said Ontology 2 node (21, 22, 23) as a match for said Ontology 1 node (11, 12, 13), when the determined causality score is lesser than the threshold, discarding (207c) the said node pair.

6. The method according to any of the claims 1-5, wherein the establishing (206) of the intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, further is based on the data associated with the respective nodes within each of Ontology 1 and Ontology 2.

7. The method according to any of the claims 1-6, further comprising:presenting (208) the result of the mapping of the ontologies, by showcasing interconnected ontologies nodes according to the similarity scores with semantic context and intra-ontology causality insights.

8. A computer program (1130) comprising instructions, which when executed by a processor (1110), causes the processor (1110) to perform actions according to any of the claims 1-7.

9. A carrier (1140) comprising the computer program (1130) of claim 8, wherein the carrier (1140) is one of an electronic signal, an optical signal, an electromagnetic signal, a magnetic signal, an electric signal, a radio signal, a microwave signal, or a computer-readable storage medium.

10. A network node (110) configured to map ontologies in a communications network (100), the network node (110) further being configured to: obtain a first ontology, Ontology 1 , adapted to represent data of a first set of data, and a second ontology, Ontology 2, adapted to represent data of a second set of data, collect a number of Ontology 1 nodes (11 , 12, 13) from the data of the first set of data, wherein each respective Ontology 1 node (11, 12, 13) comprises data from the first set of data, and a number of Ontology 2 nodes (21, 22, 23) from the data of the second set of data, wherein each respective Ontology 2 nodes (21 , 22, 23) comprises data from the second set of data, for each respective node out of said Ontology 1 nodes (11, 12, 13) and said Ontology 2 nodes (21, 22, 23), - generate a node description based on a content of the data in the node, and - embed the node description, capturing a semantic context associated with the node, determine similarity scores between the embedded Ontology 1 node (11, 12, 13) descriptions and embedded Ontology 2 node (21, 22, 23) descriptions, establish intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the Ontology 1 node (11, 12, 13) descriptions and Ontology 2 node (21, 22, 23) descriptions, and map the ontologies by mapping Ontology 1 nodes (11, 12, 13) with Ontology 2 nodes (21, 22, 23) based on the determined similarity scores and the established intra-ontology causality.

11. The network node (110) according to claim 10, wherein: the ontology mapping is adapted to relate to a data migration relating to telecommunications operations within a telecom domain, the first set of data adapted to comprise data related to a first telecommunication product, and the second set of data adapted to comprise data related to the second telecommunication product to be migrated to, the network node (110) further being configured to perform the migration from the first telecommunication product to the second telecommunication product based on the mapped ontologies.

12. The network node (110) according to any of the claims 10-11, wherein: the ontology mapping is adapted to relate to a product migration in a telecom domain, comprising subscriber migration from a first product related to handling of subscriber resource consumption, to a second product related to handling of subscriber resource consumption, the first set of data adapted to comprise data related to the first product, and the second set of data adapted to comprise data related to the second product to be migrated to, and the network node (110) further being configured to perform the subscriber migration from the first product of handling telecom operations to the second product of handling telecom operations based on said mapped ontologies.

13. The network node (110) according to claim 10, wherein: the ontology mapping is adapted to relate to mapping between a first healthcare dataset and a second healthcare dataset collected in different point of times, the first set of data adapted to comprise data related to the first healthcare dataset, and the second set of data adapted to comprise data related to the second healthcare dataset.

14. The network node (110) according to any of the claims 10-13, wherein the network node (110) further is configured to map Ontology 1 nodes (11, 12, 13) with Ontology 2 nodes (21, 22, 23) by:determine a causality score for a node pair comprising an Ontology 1 node (11, 12, 13) and an Ontology 2 node (21 , 22, 23), when the determined causality score is greater than or equal to a threshold, map the said Ontology 2 node (21 , 22, 23) as a match, for said Ontology 1 node (11 , 12, 13), when the determined causality score is lesser than the threshold, discard the said node pair.

15. The network node (110) according to any of the claims 10-14, wherein the network node (110) further is configured to establish the intra-ontology causality to identify causal relationships within each of Ontology 1 and Ontology 2, based on the data associated with the respective nodes within each of Ontology 1 and Ontology 2.

16. The network node (110) according to any of the claims 10-15, further being configured to: present the result of the mapping of the ontologies, by showcasing interconnected ontologies nodes according to the similarity scores with semantic context and intra-ontology causality insights.

Images