A method for obtaining query response data by a relational database management system (RDBMS) is provided. The method receives a user input query, by a processor associated with the RDBMS, wherein the user input query comprises a query request for a set of data; formats the user input query into a second query language suitable for communication between the RDBMS and a query response interface associated with a second data storage external to the RDBMS, by the processor, to generate a reformatted user input query, wherein the RDBMS is configured to perform query operations using an n-ary tree format, and wherein the query response interface is configured to perform query operations using a binary tree format consisting of two child nodes per non-terminal node of a binary tree; and transmits the reformatted user input query to the query response interface, via a communication device communicatively coupled to the processor.

A system includes reception of an input string of words, determination, for each subset of consecutive one or more words in the input string, of one or more phrase types based on the subset, on a dictionary describing a plurality of entities, each of the plurality of entities associated with an entity type, and on a grammar describing a plurality of phrase types, each of the plurality of phrase types associated with one or more conditions, and determination of a plurality of candidate queries based on the determined phrase types.

Technologies are described herein for executing queries expressed with reference to a structured query language against unstructured data. A user issues a structured query through a traditional structured data management (“SDM”) application. Upon receiving the structured query, an SDM driver analyzes the structured query and extracts a data structure from the unstructured data, if necessary. The structured query is then converted to an unstructured query based on the extracted data structure. The converted unstructured query may then be executed against the unstructured data. Results from the query are reorganized into structured data utilizing the extracted data structure and are then presented to the user through the SDM application.

Computer-implemented techniques for entity disambiguation using graph neural networks (GNNs) are provided. According to an embodiment, computer implemented method can comprise receiving, by a system operatively coupled to a processor, an unstructured text snippet comprising an unknown term. The method further comprises employing, by the system, a heterogeneous GNN trained on a knowledge graph associated with a domain of the unstructured text snippet to facilitate identifying one or more similar terms included within the knowledge graph for the unknown term.

The present disclosure provides a data pair generating method, apparatus, electronic device and storage medium, and the field of artificial intelligence such as natural language processing and deep learning. The method may include: generating M SQL query statements for a given database, where M is a positive integer greater than one; performing the following processing for each SQL query statement: dividing the SQL query statement into at least one SQL clause; obtaining a question description corresponding to each SQL clause; combining the question descriptions to obtain a question corresponding to the SQL query statement. The solution of the present disclosure may be applied to save manpower and time costs.

In various embodiments, methods and systems for implementing a semantic signature system are provided. A semantic signature system provides a machine trained semantic representation (i.e., a semantic signature) of the context of a word, synonyms of the word, and weak and strong relationship of the word with other words. The semantic signature can be utilized to facilitate labeling a word that is ambiguous or previously unknown. In practice, the label can be used to more accurately categorize the word for later retrieval by a search or to more accurately provide search results for a search query that includes the word.

A method may include obtaining training code and extracting features from the training code. The extracted features of the training code may be mapped to natural language code vectors by a deep neural network. A natural language search query requesting source-code suggestions may be received, and the natural language search query may be mapped to a natural language search vector by the deep neural network. The method may include mapping the natural language search query to the natural language search vector in the same or a similar method as mapping the extracted features of the training code to natural language code vectors, and the natural language search vector may be compared to the natural language code vectors. Source code responsive to the natural language search query may be suggested based on the comparison between the natural language search vector and the natural language code vectors.

In the present invention, a method for searching multilingual keywords in mixlingual speech corpus is proposed. This method is capable of searching audio as well as text keywords. The capability of audio search enables it to search out-of-vocabulary (OOV) words. The capability of searching text keywords enables it to perform semantic search. An advanced application of searching keyword translations in mixlingual speech corpus is also possible within posteriorgram framework with this system. Also, a technique for combining information from text and audio keywords is given which further enhances the search performance. This system is based on multiple posteriorgrams based on articulatory classes trained with multiple languages.

A computer-implemented method includes: receiving, by a computing device, an input file defining correct spellings of one or more transliterated words; generating, by the computing device, suffix outputs based on the one or more transliterated words; generating, by the computing device, a dictionary that maps the suffix outputs to the one or more transliterated words; recognizing, by the computing device, an alternatively spelled transliterated word included in a document as one of the one or more correctly spelled transliterated words using the dictionary; and outputting, by the computing device, information corresponding to the recognized transliterated word.

Methods, systems, and computer programs are presented for improving search speed and quality using differential indexing. One method includes an operation for building a first index for a database, the first index being for first tokens resulting from normalizing words in input data. Further, the method includes building a second index for the database, the second index being for second tokens comprising words of the input data eliminated from the first index during the normalizing. The method further includes operations for receiving a raw query for a search of the database, and for generating a search query based on tokens of the raw query. The search query comprises a combined search of the first index and the second index. A search is performed based on the search query, and results of the search are returned for presentation on a display.