Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Acknowledgement
Acknowledgement is made of applicant’s amendment made on 06/12/2026. Applicant’s submission filed has been entered and made of record.
Status of the Claims
Claims 1-20 are pending.
Response to Applicant’s Arguments
In response to “However, the subject of rearrangement in Yoon is a plurality of sentences within a single utterance. Yoon merely discloses classifying a single user voice into multiple sentences and identifying the intent included in each sentence. See Yoon at paragraph [0058]. This sentence- level classification is merely a division of a single utterance input by the user at one time. Yoon does not disclose extracting intents respectively from a plurality of independent utterances as recited in Claims 1 and 11. In contrast, the first utterance and the second utterance of claim 1 are temporally independent utterances, as supported, for example, by paragraph [0049] of the present specification, which states that "[t]he second utterance is performed after the first intent for the first utterance is extracted." Specification at paragraph [0049]”.
Yoon teaches “When it is identified that a plurality of intents are included in a user voice, the processor may obtain word use information from a sentence including the plurality of intents at operation S310” (¶57). Specifically, while the processor may identify a user's intent included in each sentence by dividing the user voice in units of sentences, but is not limited thereto (¶58).
Further, Yoon does not define what is considered a “single utterance” or “multiple utterance”. Neither does the specification.
Broadly construed, “if a user voice such as “I think my smartphone is defective. I can't user may phone because it's hot. Can I exchange it for free?” is input, the processor may classify the user voice as 3 pieces of data, that is, “I think my smartphone is defective”, “I can't use the phone because it's hot” and “Can I exchange it for free?.”” (¶58), the user voice comprises three pieces, each can be an utterance, one after the other.
Therefore, Yoon is not limited to rearrange intents corresponding to sentences within a single utterance.
In response to “Additionally, the purpose of the rearrangement in Yoon is merely to determine a response order. Specifically, in Yoon, the rearrangement merely adjusts a priority order among a plurality of intents, and its output is nothing more than a response order. See Yoon at paragraph [0067]. In contrast, the rearrangement recited in claim 1 is an intermediate process of generating a final intent, wherein rearranged partial intents are combined to generate a new final intent. See, e.g., Specification at paragraph [0090] ("Referring to FIG. 5, the processor 120 may extract partial intents based on a plurality of utterances and may rearrange the partial intents. The process of rearranging partial intents may be understood as an intermediate process of generating final intents."). Applicant respectfully asserts that the rearrangement of Yoon, which simply reorders already-identified intents to determine a response order, is clearly distinct from the rearrangement of claim 1, which is directed to generating a new final intent. As amended, independent claim 1 clarifies that the second utterance is temporally independent from and obtained after the first utterance. As discussed above, neither Arizmendi”.
“Since the response order corresponding to each sentence is determined according to the priority of the sentences, the sentences with the same or similar intent may be set as the adjacent priority so that the answer to the same or similar intent may be continuously provided. For example, if #2 is a battery problem, #1 is a battery consumption problem, the answers regarding the battery may be provided continuously by placing #2 and #1 adjacently. Thus, the understanding and convenience of the user may be increased” (¶70).
In other words, Yoon teaches answering the same or similar intent continuously: if the intent #1 is the same as intent #2, then the answer to intent #1 is the same as the answer to intent #1.
Further, “Specifically, the fourth model may identify “parent intent_id” of each sentence and group sentences including the same “parent intent_id.”” (¶112). As such, Yoon teaches generating a final intent “parent intent_id” for a plurality of same intents.
Claim Rejections - 35 USC § 103
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 103 that form the basis for the rejections under this section made in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-7 and 11-17 are rejected under 35 USC 103(a) as being unpatentable over Marzinzik (US 11705114 B1) in view of Arizmendi et al. (US 2014/0156268 A1).
Regarding Claims 1 and 11, Marzinzik discloses a voice recognition apparatus (Fig. 4, User Computer Device) comprising:
a microphone configured to extract an utterance of a user (Col 9, Rows 13-14, user computer device in communication with a microphone);
a memory (Col 10, Rows 58-61, memory 410 stores and retrieves executable instructions) configured to store a scenario matching an intent extracted from the utterance (Col 10, Rows 5-16, orchestrator 312 as part of executing app 310 and code 307 with access to one or more databases 320 corresponding to an orchestrator model that knows intents of a domain per Col 6, Rows 45-46); and
a processor (Col 10, Rows 56-57, processor 405) configured to:
search for the scenario based on the utterance (Col 6, Rows 30-47, send utterance to an orchestrator model that knows the intents of the domain; i.e., per Fig. 1, steps 112 and 124, in a multi-turn conversation, select orchestrator model that knows the intents of the domain for processing utterance in a current turn);
perform a voice recognition function (Col 6, Rows 30-47, translate audio stream 106 from speech to text, determine a long pause being end of a statement, and process the statement using the selected orchestrator model);
extract a first intent from a first utterance (Col 6, Rows 45-47, at the end of the first statement, process the first statement to identify top intent by sending the utterance to the orchestrator);
extract a second intent from a second utterance obtained after the first utterance (Col 6, Rows 55-57, if pause is the end of user’s talking, process user’s statements corresponding to user’s turn; i.e., subsequent to processing the first statement in Col 6, Rows 40-42, process a second statement obtained after the first statement per Col 6, Rows 55-57);
separate the first intent into first partial intents / intent units (Col 6, Rows 40-42 and 50, a statement includes multiple intents such that processing the first statement to identify the top intent identifies multiple intents; e.g., Fig. 12 and Col 14, Rows 40-56, split “Tom and Jerry are going and so are Sally and Jim” into “Tom and Jerry are going” and “and so are Sally and Jim”, and divide the verbal statement into a plurality of intents based upon the splits per Col 16, Rows 13-17) and the second intent into second partial intents / intent units by using separators (Col 6, Rows 55-57 and Col 15, Rows 54-67, when processing second statement in the user’s statements, e.g., split “He went to the store with Jim” into “He went to the store” and “with Jim”, and divide the verbal statement into a plurality of intents based upon the splits per Col 16, Rows 13-17);
rearrange the first partial intents / intent units of the first intent and the second partial intents / intent units of the second intent (Col 6, Rows 58-64, sort and prioritize all of the intents to process all of the intents in proper order); and
generate a final intent by combining the first partial intents / intent units of the first intent and the second partial intents / intent units of the second intent (Col 6, Rows 60-61 and Claim 1, determine a priority of each of the plurality of intents and process the intents in an order corresponding to the determined priority of each intent; i.e., the final intent comprises prioritized partial intents sorted in proper order), such that duplicate partial intents are deleted depending on definitions of the separators,
perform the voice recognition function based on the scenario matching the final intent (Col 5, Rows 6-18, perform actions based upon user’s requests when placing a phone call to order a pizza per Fig. 14 by selecting an orchestrator model in the food ordering domain to process statements of Fig. 14, process all of the intents in proper order per Col 6, Rows 60-63, and generate the response based on the intents per Col 16, Rows 15-18 to);
wherein each of the first partial intents / intent units and the second partial intents / intent units indicates at least one of an action (Col 5, Rows 51-53, determine a response to the user based upon the statements made by the user), a target (Fig. 14, responding to “I need to order three pizzas with cheese and pepperoni and have them delivered to 123 Main Street”), or an entity (Col 5, Rows 36-37, extract entities from the identified intents) included the first intent or the second intent.
Marzinzik does not teach combining the first partial intents of the first intent and the second partial intents of the second intent such that duplicate partial intents are deleted depending on definitions of the separators.
Arizmendi discloses a voice recognition apparatus (Fig. 1) performing a voice recognition function (¶35, integrating incremental speech recognition results with a partially observable Markov decision process (“POMDP”) dialog manager (“DM”) requires using an incremental automatic speech recognition (“IASR”) module 202 per ¶¶28-29, which analyzes speech input and provide incremental speech recognition results / incremental textual transcriptions of the speech input as output);
extracting intent from the utterance (¶29, SLU module 204 can receive the transcribed input and can use a natural language understanding model to analyze the group of words that are included in the transcribed input to derive a meaning from the input);
separating the intent into partial intents by using separators (¶30, SLU 204 processing partial results from IASR 202; ¶36, recognized text from the incremental speech recognition result may be incomplete and may end in the middle of a sentence or even word; i.e., SLU 204 separately generates partial intents for respective partial ASR results ending in the middle of a sentence or word); and
generating a final intent by combining partial intents, such that duplicate partial intents are deleted depending on definitions of the separators (¶29, dialog manager DM 206 receives meaning of the speech input from SLU 204 and determines an action; ¶34, Table I, based on a determination (i.e., definition) that the fourth action indicated by a first partial intent and the fifth action indicated by a second partial intent are the same (i.e., duplicates), reject the fifth action; in view of ¶¶16-17, if an action is rejected (e.g., duplicate fifth action), then the copy of the dialog manager state (per ¶19, a belief state incorporating contextual information available to the system from the beginning of the dialog) corresponding to the rejected action is discarded (i.e., deleted)).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to combine the first partial intents of the first intent and the second partial intents of the second intent such that duplicate partial intents are deleted depending on definitions of the separators such that an intent / action that does not sufficiently advance a dialog, it is rejected and discarded (Arizmendi, ¶31).
Regarding Claim 2, Marzinzik discloses wherein the processor is further configured to generate the final intent based on the first utterance and the second utterance in response to one of or any combination of the action, the target, or the entity, being missing from the second utterance obtained after the first utterance (Col 6, Rows 55-66, at the end of user’s turn / user’s talking, determine if there are any missing entities when processing all of the intents in proper order, and to request the missing entities from the user).
Regarding Claims 3 and 13, Marzinzik discloses wherein the processor is further configured to separating the first intent and the second intent into the first partial intent units and the second partial intent units comprising: assign one separator among the separators to each one of or any combination of the action (Col 13, Rows 31-45 and Fig. 8, determine a split for “I want to extend my stay” and “for my room number one”; see also Fig. 14, Col 5, Rows 12-18 and Col 15, Rows 30-33, determining split to order pizza action), the target (Fig. 13, Col 13, Row 66 – Col 15, Row 9, parse the statement “I need to buy a fuse for a number of 1 2 3” based on a preposition or subordinating conjunction “for” and “of” and place final split before the word “for a number of 1 2 3”), and the entity (Col 15, Rows 19-22, Fig. 14, check each side of coordinating conjunctions for nouns to determine a split; per Col 6, Rows 50-52, extract the entities from the identified intents), included in each of the first intent and the second intent (Col 7, Rows 10-18, break up compound and complex statements into smaller utterances for intent recognition to determine if they are intents; e.g., Col 4, Rows 24-30, break up “I want to extend my stay for my room number abc” into “I want to extend my stay” and “for my room number abc” to determine intents).
Regarding Claims 4-5 and 14-15, Marzinzik discloses wherein the processor is further configured to separating the first intent into the first partial intent units and the second intent into the second partial intent units comprising:
assign a first separator to the first partial intents included in the first intent, wherein the first separator points to the action or the target included among the first partial intents extracted from the first intent (Col 6, Rows 50-54, at the end of a statement including multiple intents, identify all the intents); and
assign a second separator to the second partial intents included among the second partial intents extracted from the second intent (Col 6, Rows 55-62 and Col 16, Rows 15-17, at the end of user’s turn / talking, process user’s statements to detect one or more potential splits in the verbal statements and divide the verbal statements into a plurality of intents based on the potential splits).
Marzinzik does not disclose wherein the second separator points to the action or the target being opposite to the first partial intents to which the first separator was assigned.
Arizmendi discloses wherein the processor is further configured to:
assign a first separator to the first partial intents, wherein the first separator points to the action or the target included among the first partial intents extracted from the first intent (¶30 and ¶36, SLU 204 processing partial results ended in the middle of a sentence from IASR 202 and providing results to IIM 212 (i.e., SLU 204 separately generates partial intents for respective partial ASR results ending in the middle of a sentence); ¶¶31-32, IIM 302 copies the current state of dialog manager, provides copied temporary instance of dialog manager with an incremental speech recognition result, and inspects the action that the copied dialog manager would take); and
assign a second separator to the second partial intents included among the second partial intents extracted from the second intent (¶32, provide a second copy of the original dialog manager with a new incremental speech recognition result 304 and determine if the second copy takes an action that advances the dialog),
wherein the second separator points to the action or the target being opposite to the first partial intents to which the first separator was assigned (¶31, when inspecting the action that the copied dialog manager with the incremental speech recognition result 304 would take, determine that the action does not sufficiently advance the dialog that would require re-asking the same question; ¶32, make a second copy of the original dialog manager and a new incremental speech recognition result that is a revision of the previous incremental speech recognition result, evaluate or determine that the second copy takes an action that advances to the dialog; i.e., the first action does not advance the dialog being opposite to second action that does advance the dialog);
wherein the processor is further configured to generate the final intent by deleting a particular first partial intent and a particular second partial intent that are the same as each other, in response to the particular first partial intent to which the first separator is assigned being the same as the particular second partial intent to which the second separator is assigned (¶34, begin to execute the action generated by the third partial “Mckee” and the fourth partial “Mckeesport” revises the action; per ¶33, incomplete incremental speech recognition (e.g., “Mckee”) can be discarded; see also ¶34, fifth partial “Mckeesport Center” where corresponding fifth action is rejected since it is the same; in view of ¶¶16-17, if an action is rejected (e.g., duplicate fifth action), then the copy of the dialog manager state (per ¶19, a belief state incorporating contextual information available to the system from the beginning of the dialog) corresponding to the rejected action is discarded (i.e., deleted)).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to assign a second separator to the second partial intents included among the second partial intents extracted from the second intent wherein the second separator points to the action or the target being opposite to the first partial intents to which the first separator was assigned interact with SLU processing partial results to provide partial intents at any point or at multiple points (Arizmendi, ¶30).
Regarding Claims 6-7 and 16-17, Marzinzik discloses wherein the processor is further configured to separating the first intent into the first partial intent units and the second intent into the second partial intent units comprising:
assign a third separator to a first entity included among the first partial intents / intent units in the first intent (Col 6, Rows 50-52, extract the entities from the identified intents; Col 15, Rows 19-22, Fig. 14, check each side of coordinating conjunctions for nouns to determine a split); and
assign a fourth separator to a second entity included among the second partial intents / intent units in the second intent (Col 6, Rows 55-57, process user’s statements including extracting entities from the identified intents of user’s statements including second statement),
Marzinzik does not disclose wherein the fourth separator negates the first entity to which the third separator was assigned.
Arizmendi discloses wherein the processor is further configured to:
assign a third separator to a first entity included among the first partial intents and assign a fourth separator to a second entity included among the second partial intents, wherein the fourth separator negates the first entity to which the third separator was assigned (¶32, if the second copy of the dialog manager (which is associated with the new ISR that is a revision; e.g., entity “Austin” in the previous ISR result was revised to entity “Boston” in the revised ISR result) takes an action that advances the dialog and is different from the action generated by the first copy, then terminate the first action, discard the first copy of the dialog manager, initiate the second action, and the second copy assumes the position of the first copy; i.e., when generating a copy of the dialog manager for respective incremental speech recognition (i.e., first separator for the first ISR per ¶31 and second separator for the new ISR per ¶32), generate a first position (third separator) for the copied state of dialog manager and corresponding ISR (entity “Austin”) and a second position (fourth separator) for the copied state of dialog manager and corresponding revised ISR (entity “Boston”), the copied state of dialog manger and corresponding revised ISR “Boston” takes the position of the copied state of the dialog manager and corresponding ISR “Austin”); and
generate the final intent by deleting the entities that are the same as each other (in another example, see ¶34 and Table I, reject incremental speech recognition result “Mckeesport center” since it is the same as incremental speech recognition result “Mckeesport”; in view of ¶¶16-17, if an action is rejected (e.g., duplicate fifth action), then the copy of the dialog manager state (per ¶19, a belief state incorporating contextual information available to the system from the beginning of the dialog) corresponding to the rejected action is discarded (i.e., deleted)) and for which the first entity to which the third separator is assigned is the same as the second entity to which the fourth separator is assigned (¶34 and Table I, third separator being position assigned to the copied dialog manager state associated with incremental speech recognition result “Mckeesport” and fourth separator being position assigned to the copied dialog manager state associated with incremental speech recognition result “Mckeesport center”, which refers to the same entity “Mckeesport”)
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to assign a third separator to a first entity included among the first partial intents and assign a fourth separator to a second entity included among the second partial intents, wherein the second entity to which the fourth separator was assigned negates the first entity to which the third separator was assigned in order to gracefully handle revisions following a premature action execution (Arizmendi, ¶31).
Regarding Claim 12, Marzinzik discloses wherein extracting the second intent from the second utterance is performed after extracting the first intent from the first utterance (Col 6, Rows 40-50, if the pause is end of statement (i.e., first statement), process the first statement to identify corresponding intent or multiple intents (i.e., first partial intents / intent units); Col 6, Rows 55-62, if pause is end of user’s talking / turn, process user’s statements (i.e., multiple statements including a subsequent / second statement), process user’s statements to identify corresponding intents include second partial intents / intent units of the subsequent second statement).
Claims 8-10 and 18-20 are rejected under 35 USC 103(a) as being unpatentable over Marzinzik (US 11705114 B1) and Arizmendi et al. (US 2014/0156268 A1) as applied to claims 1 and 11, in view of Yoon et al. (US 2021/0166687 A1).
Regarding Claims 8 and 18, Marzinzik does not disclose wherein the processor is configured to rearrange the first partial intents derived from the first utterance and the second partial intents derived from the second utterance in a reverse order of utterances.
Yoon discloses a voice recognition apparatus comprising a processor performing a voice recognition function (Fig. 1, server 200 including a processor per ¶54; ¶59, processor converts user voice into text data through an ASR model), extracting a first intent from a first utterance (¶58, processor divides user voice in units of sentences to identify a user’s intent included in each sentence), extracting a second intent from a second utterance (¶58, processor divides user voice in units of sentences to identify a user’s intent included in each sentence), and rearrange the first intent and the second intent (¶67, distinguish user’s voice in units of sentences from #1 to #5, each sentence includes an intent where the processor first identify the priority of each sentence / intent in the order of #2, #5, #1, #3, and #4; ¶¶68-69, rearrange the first priority based on intent information in the order of #2, #1, #5, #3, and #4);
wherein the processor is configured to rearrange the first partial intents derived from the first utterance and the second partial intents derived from the second utterance in a reverse order of utterances (Yoon, ¶¶68-69, rearrange the first priority #2, #5, #1, #3, and #4 based on intent information in the order of #2, #1, #5, #3, and #4; i.e., #5 and #1 have been reversed in the priority order).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to divide user utterances into units of sentences to extract a first intent from a first sentence / utterance (including first partial intents per established function of Marzinzik at Col 16, Rows 15-17), a second intent from a second sentence / utterance (including second partial intents per established function of Marzinzik at Col 16, Rows 15-17), and rearrange the first intent and the second intent in a reverse order of utterances so that same or similar intents become adjacent in the priority order (Yoon, ¶68).
Regarding Claims 9 and 19, Marzinzik as modified by Arizmendi to gracefully handle revision (Arizmendi, ¶32) discloses wherein the processor is configured to generate the final intent by deleting actions other than a most preceding action among the first partial intents and the second partial intents (Arizmendi, ¶34 and Table I, discard or reject actions corresponding to partial results “Yew”, “Ridge”, as well as action for “Mckee” since it is incomplete per ¶33; execute revised action for “Mckeesport”).
Regarding Claims 10 and 20, Marzinzik as modified by Arizmendi to gracefully handle revision (Arizmendi, ¶32) discloses wherein the processor is configured to generate the final intent by deleting particular partial intents that match deleted partial intents, from among the first partial intents and the second partial intents (Arizmendi, ¶34, fifth action for fifth partial “Mckeesport center” is rejected since it is same as the third partial “Mckee”, which was discarded as being incomplete result of fourth partial “Mckeesport”).
Conclusion
Applicant's amendment necessitated the new grounds of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to examiner Richard Z. Zhu whose telephone number is 571-270-1587 or examiner’s supervisor Hai Phan whose telephone number is 571-272-6338. Examiner Richard Zhu can normally be reached on M-Th, 0730:1700.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/RICHARD Z ZHU/Primary Examiner, Art Unit 2654 07/29/2026
Prosecution Insights