DETAILED ACTION
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 .
Examiner Notes on Patent Subject Matter Eligibility Under 35 USC § 101
Claims 1-21 are directed to patent-eligible subject matter under 35 U.S.C 101. Although the claims include language relating to spoken language understanding, intent detection, slot filling, vectors, matrices, encoders, attention weights, and loss functions, the claims are not merely directed to organizing information, comparing text, or performing mental steps in the abstract. Rather the claims integrate such processing into a practical application within a spoken language understanding system by using a particular neural-network architecture including a task-shared encoder, intent-specific encoder, slot-specific encoder, label attention component, intent-slot co-attention component, multiple intent decoder, slot decoder, and objective-loss based training process.
The claimed operations are not capable of being practically performed in the human mind because the claims require generating and processing high dimensional neural network representations, including word token vectors, intent-specific latent vectors, and more. The claims encoders and decoders perform machine-learning operations that are part of a trained neural network SLU architecture rather than generic human observation or judgement. The specification and cited art demonstrate that SLU systems use neural network components such as BiLSTM encoders, self-attention modules, label-attention matrices, heterogenous graph/label interaction layers, and decoders to improve intent recognition and slot filling in spoken dialogue systems.
The claimed subject matter further integrates any mathematical operations into a practical application because the vectors, matrices, attention computations, and loss functions are not claimed din isolation. They are used to train and operate a SLU system that receives an utterance, generates task shared and tasks specific representations, models relationships between intent labels and slot labels, etc. The claims therefore apply the mathematical computations as part of a concrete SLU training and prediction architecture for improving the performance of intent recognition and slot filling.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth 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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 2, 11, 12, and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (hereinafter Zhu) (CN 117349439 A) (see attached copy for paragraph numbers) in view of Vu et al. (hereinafter Vu) (A Label Attention Model for ICD Coding from Clinical Text).
Regarding claim 1, Zhu discloses:
A method for training of a spoken language understanding (SLU) system comprising (Zhu, P[0001], SLU, P[0093], training, P[0107], learning/optimizes):
obtaining, by a task-shared encoder, an input utterance consisting of n word tokens w1,w2,…,wn (Zhu, P[0045]: "For an input utterance U, hidden state representations for intent recognition and slot filling tasks are obtained", "where n is the utterance length and d is the hidden state dimension." (teaches obtaining an input utterance U having utterance length N, which reads on an input utterance consisting of n word tokens.));
generating, by the task-shared encoder, a vector ei to represent an ith word token wi based on the word tokens w1,w2,…,wn (Zhu, P[0009]: "a self-attention encoder is first used to obtain the shared utterance representation of two sub-tasks, wherein the input U is sequentially fed into a bidirectional long short-term memory network and a self-attention mechanism module; then the two word representation streams are connected as the task-shared utterance representation E" (teaches a task-shared encoder that processes the input U through BiLSTM and self-attention modules and connects word representation streams as task-shared utterance representation E, reading on vectors ei representing word tokens wi.));
generating, by an intent-specific encoder, intent-specific latent vectors eiI for intent detection based on a sequence of vectors e1:n, wherein the intent-specific latent vectors are concatenated to formulate an intent-specific matrix EI (Zhu, P[0009]: "subsequently, two different bidirectional long short-term memory networks are used on E to obtain the hidden state representation of the intent recognition task and the hidden state representation of the slot filling task" (teaches using one of two different BiLSTMs on the task-shared utterance representation E to obtain hidden state representations for the intent recognition task, reading on intent-specific latent vectors based on e1:n and concatenated/formulated as EI.));
generating, by a slot-specific encoder, slot-specific latent vectors eiS for slot filling based on the sequence of vectors e1:n, wherein the slot-specific latent vectors are concatenated to formulate a slot-specific matrix ES (Zhu, P[0009]: "subsequently, two different bidirectional long short-term memory networks are used on E to obtain the hidden state representation of the intent recognition task and the hidden state representation of the slot filling task" (teaches using one of two different BiLSTMs on the task-shared utterance representation E to obtain hidden state representations for the intent recognition task, reading on intent-specific latent vectors based on e1:n and concatenated/formulated as ES.));
generating, by a slot-specific encoder, an intent label-specific matrix VI (Zhu, P[0046]: "For the input utterance, obtain the hidden state representation of the intent and slot label.", P[0048]: "Uses discourse self-attention mechanism to capture label-related contextual information from discourse as a representation of intent and slot labels" (teaches obtaining intent label representations from the utterance, reading on an intent label-specific matrix VI.))
generating, by the label attention component, slot label-specific matrices VS,k (Zhu, P[0046]: "For the input utterance, obtain the hidden state representation of the intent and slot label.", P[0048]: "Uses discourse self-attention mechanism to capture label-related contextual information from discourse as a representation of intent and slot labels" (teaches obtaining slot label representations from the utterance, reading on an slot label-specific matrices Vs,k.))
generating, by an intent-slot co-attention component, a multiple intent representing matrix H←1 based on the slot-specific matrix ES (Zhu, P[0008]: "For the input utterance, obtain the hidden state representation for the intent recognition task and the slot filling specific task;" (Zhu teaches slot-filling specific hidden state representation corresponding to ES.) the intent label-specific matrix VI and the slot label-specific matrices VS,k (Zhu, P[0008]: "S2. For the input utterance, obtain the hidden state representation of the intent and slot label;" (Zhu teaches intent and slot label representations corresponding to VI and VS,k., Zhu, P[0008]: "S3. The dual heterogeneous graph interaction layer is used to capture statistical dependencies within and between labels in the training corpus and to enhance the interactions between labels for each utterance." (Zhu teaches an interaction layer enhancing interactions between intent/slot labels for each utterance, reads on generating enhanced multiple intent representing matrix based on ES, VI, and VS,k.));
generating, by the intent-slot co-attention component, a slot representing matrix H→l+2 based on the slot-specific matrix ES (Zhu, P[0008]: "For the input utterance, obtain the hidden state representation for the intent recognition task and the slot filling specific task" (teaches slot-filling specific hidden state representation corresponding to ES.)), the intent label-specific matrix VI and the slot label-specific matrices VS,k (Zhu, P[0008]: "For the input utterance, obtain the hidden state representation of the intent and slot label" (teaches intent and slot label representations corresponding to VI and VS,k.), Zhu, P[0013]: "S3.2 enhancing the interaction between labels of each utterance through the dynamic local heterogeneous graph interaction layer to update the label representation." and P[0018]: "Integrate label information and contextual features and add them to Eδ to obtain enhanced label information:" (teaches updating/enhancing label representations and integrating label information and contextual features, reading on generating a slot representing matrix based on ES, VI, and VS,k.));
generating, by a multiple intent decoder, final intent outputs based on the multiple intent representing matrix H1← (Zhu, P[0078]: "Using a spoken language understanding decoder, the intent and slot label of the current utterance are predicted." (teaches using an SLU decoder to predict intent labels.)) and the intent label-specific matrix VI (Zhu, P[0087]-P[0089]: "
Ultimately, I was used for multi-intent recognition: The predicted sentence-level intent o<sup>I</sup> is obtained through a voting mechanism." (teaches generating final multi-intent outputs based on enhanced intent/label information.));
generating, by a slot decoder, final slot output based on the slot representing matrix H→l+2 (Zhu, P[0078]: "Using a spoken language understanding decoder, the intent and slot label of the current utterance are predicted." (teaches using an SLU decoder to predict slot labels.) and the slot-specific matrix ES (Zhu, P[0090]-P[0092]: "Similarly, S is used for slot filling: Finally, the output o<sup>S</sup> = argmax(S) is the predicted slot sequence of U." (teaches generating a final slot output/predicted slot sequence based on slot information.));
computing, by the multiple intent decoder, an intent detection loss LID (Zhu, P[0094]: "Specifically, the loss function for multi-intent recognition is defined as: " (teaches computing a multi intent recognition loss, reading on intent detection loss LID.));
computing, by the slot decoder, a slot filling loss LSF (Zhu, P[0096]: The loss function for slot filling is defined as:" (teaches computing slot filling loss LSF.)); and
adjusting the SLU system based on an objective loss L, wherein the objective loss L is a weighted sum of the intent detection loss LID and the slot filling loss LSF (Zhu, P[0098]-P[0100]: "The final loss function is the sum of the two: Where Θ represents all trainable parameters, and λ is a tradeoff hyperparameter." (teaches final loss function combining the multi-intent recognition loss and slot filling loss with Θ representing trainable parameters and λ as a tradeoff hyperparameter, reading on adjusting the SLU system based on weighted objective loss.)).
Zhu does not explicitly disclose:
based on the intent-specific matrix EI
based on the slot-specific matrix ES
However, Vu discloses:
based on the intent-specific matrix EI (Vu, Page 3: "Our label attention mechanism takes H as the input and output jLj label-specific vectors representing the input document D.", Equation 6: "V = HA^T" (teaches generating label-specific vectors/matrix V from an encoded hidden-state matrix H; applying Vu's label-attention mechanism to Zhu's intent-specific hidden-state representation will yield VI based nn EI.));
based on the slot-specific matrix ES (Vu, Page 3: "Our label attention mechanism takes H as the input and output jLj label-specific vectors representing the input document D.", Equation 6: "V = HA^T" (teaches generating label-specific vectors/matrix V from an encoded hidden-state matrix H; applying Vu's label-attention mechanism to Zhu's slot-specific hidden-state representation will yield Vs,k based on ES.));
It would have been prima facie obvious to one of ordinary skill in the art before the earliest filing date of the claimed invention to have modified Zhu in view of Vu. Doing so would have provided Zhu’s multi-intent spoken language understanding method, which obtains task shared utterance representations, intent/slot hidden state representations, and enhances interactions between labels for each utterance (Zhu, P[0008]-P[0013]), with Vu’s known label attention mechanism that takes hidden state matrix H as input and computes label attention using specific formulas (Vu, Page 3), thus, allowing Zhu’s SLU system to generate label-specific intent and slot matrices from intent/slot hidden representations to improve intent recognition and slot filling.
Regarding claim 2, the combination of Zhu and Vu discloses the method of claim 1.
Zhu further discloses:
receiving, by an audio input device, the input utterance (Zhu, P[0002]: "Task-oriented dialogue systems have gained widespread attention due to the extensive use of intelligent voice assistants such as Apple Siri and Microsoft Cortana." (teaches SLU method in the context of intelligent voice assistants, which conventionally receive spoken input utterances through an audio input device), "As an intermediary interface between users and machines, spoken language comprehension plays a crucial role in task-oriented dialogue systems." (teaches SLU as an interface between uses and machines in task-oriented dialogue systems, reads on receiving user input utterance for SLU processing)); and
generating, by an output device, an output utterance based on the input utterance (Zhu, P[0002]: "As an intermediary interface between users and machines, spoken language comprehension plays a crucial role in task-oriented dialogue systems." (teaches SLU in a user-machine dialogue interface, where generating an output utterance based on the understood input utterance would have been an obvious and conventional dialogue-system output.), P[0034]: "The method of this invention is scalable and transferable, and can be applied to other multi-tasks in dialogue systems, such as joint action recognition and sentiment analysis. " (confirms method is applied in dialogue systems, which conventionally generate output responses based on received user utterances)).
Regarding claim 11, claim 11 recites the spoken language (SLT) system corresponding to the methods described in claim 1 and is rejected under the same grounds as above.
Zhu further recites:
A spoken language understanding (SLU) system (Zhu, Title, Abstract) comprising:
one or more processors (Zhu, P[0121]); and
a computer-readable medium having instructions stored there on, which, when executed by the one or more processors, cause the system to perform operations comprising (Zhu, P[0121]):
Regarding claim 12, claim 12 recites the spoken language (SLT) system corresponding to the methods described in claim 2 and is rejected under the same grounds as above.
Regarding claim 21, claim 21 recites the non-transitory storage medium corresponding to the methods described in claim 1 and is rejected under the same grounds as above.
Zhu further recites:
A non-transitory computer-readable storage medium comprising instructions that, when executed by at least one processor of a machine (Zhu, P[0121]).
Claims 3-6, and 13-16 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (hereinafter Zhu) (CN 117349439 A) (see attached copy for paragraph numbers) in view of Vu et al. (hereinafter Vu) (A Label Attention Model for ICD Coding from Clinical Text) in further view of Lample et al. (hereinafter Lample) (Neural Architectures for Named Entity Recognition).
Regarding claim 3, the combination of Zhu and Vu discloses the method of claim 2.
Zhu further discloses:
wherein the generating the vector ei to represent the ith word token wi comprises concatenating a contextual word embedding eiBiLSTMword (Zhu, P[0009]: "the input U is sequentially fed into a bidirectional long short-term memory network " (teaches BiLSTM-based word representation stream, reading on the contextual word embedding eiBILSTMword.)), a self-attention embedding eiSA (Zhu, P[0009]: "and a self-attention mechanism module" (teaches a self-attention-based word representation stream, reading on the self-attention embedding eiSA.))
which is according to the following formula:
ei= eiBiLSTMword ⊕eiSA (Zhu, P[0009]: "then the two word representation streams are connected as the task-shared utterance representation E" (teaches connecting/concatenating the BiLSTM based and self-attention based word))
wherein:
the contextual word embedding eiBiLSTMword is an embedding of the token word wi, derived by applying a single bidirectional BiLSTM layer (Zhu, P[0009]: "the input U is sequentially fed into a bidirectional long short-term memory network" (teaches applying a BiLSTM to derive a word/token representation.))
the self-attention embedding eiSAis an embedding of word wi, derived by applying a single self-attention layer (Zhu, P[0009]: "a self-attention encoder is first used to obtain the shared utterance representation of two sub-tasks" (teaches deriving a self-attention word/utterance representation)))
Zhu does not explicitly disclose:
and a character-level word embedding ewiBiLSTMchar.
⊕ewiBiLSTMchar.
to real valued embedding representations ew1,ew2,…,ewn
to the real valued embedding representations ew1,ew2,…,ewn
and
the character-level word embedding ewiBiLSTMchar.is a character-level word embedding, derived by applying another single BiLSTM (BiLSTMchar.) to the real valued embedding representations of characters in each word token wi.
However, Vu further discloses:
to real valued embedding representations ew1,ew2,…,ewn (Vu, page 3: " vectors ew1, ew2, ...,ewn" (teaches sequence of word embedding vectors ew1, ew2, …, ewn));
to the real valued embedding representations ew1,ew2,…,ewn (Vu, page 2: "ith token w", page 3: " vectors ew1, ew2, ...,ewn" (teaches sequence of word embedding vectors ew1, ew2, …, ewn));
Vu, in combination with Zhu, does not explicitly disclose:
and a character-level word embedding ewiBiLSTMchar.
⊕ewiBiLSTMchar.
and
the character-level word embedding ewiBiLSTMchar.is a character-level word embedding, derived by applying another single BiLSTM (BiLSTMchar.) to the real valued embedding representations of characters in each word token wi.
However, Lample discloses:
and a character-level word embedding ewiBiLSTMchar. (Lample, Page 6: "The embedding for a word derived from its characters is the concatenation of its forward and backward representations from the bidirectional LSTM" (teaches character-level BiLSTM word embedding))
⊕ewiBiLSTMchar. (Lample, Page 6: "This character-level representation is then concatenated with a word-level representation from a word lookup-table" (teaches concatenating character-level representation with word level representation, yields claimed three part concatenation));
and
the character-level word embedding ewiBiLSTMchar.is a character-level word embedding (Lample, Page 6: "character-level representation" (teaches a character-level word representation), derived by applying another single BiLSTM (BiLSTMchar.) (Lample, Page 2: "This forward and backward LSTM pair is referred to as a bidirectional LSTM" (teaches applying a character-level BiLSTM)) to the real valued embedding representations of characters in each word token wi (Lample, Page 6: "embedding for every character" (teaches real-valued character embeddings for characters in each word)).
It would have been prima facie obvious to one of ordinary skill in the art before the earliest filing date of the claimed invention to have modified Zhu in view of Vu and in further view of Lample. Doing so would have provided Lample’s known character level BiLSTM representation in which character embeddings are provided to forward and backward LSTMs and the resulting word representations is the concatenation of forward and backwards representations (Lample, Page 2) with Zhu’s multi-intent spoken language understanding method, which obtains task shared utterance representations, intent/slot hidden state representations, and enhances interactions between labels for each utterance (Zhu, P[0008]-P[0013]), and Vu’s known label attention mechanism that takes hidden state matrix H as input and computes label attention using specific formulas (Vu, Page 3), thus, allowing Zhu’s SLU system to include character-level word information in its token representations to improve sequence labeling and slot filling, including for are or unseen words.
Regarding claim 4, the combination of Zhu, Vu, and Lample discloses the method of claim 3.
Vu further discloses:
wherein the generating, by the label attention component, the intent label-specific matrix VIbased on the intent-specific matrix EI, comprises:
computing an intent label-specific attention weight matrix AI based on a following formula: AI=softmax(BI×tanhDI×EI) (Vu, Equations 4 and 5, Page 3, (teaches computing a label-specific attention weight matrix using the same structure, where H maps ot EI, W maps to DI, U maps to BI, and A maps to AI));
wherein the softmax is performed at a row level to make sure that a summation of weights in each row is equal to 1 (Vu, Page 3: "row level" and "summation of weights" (teaches row-level softmax normalization)), and ΒI∈RLI×da (Vu, Page 3: "U ∈ R|L|×da" (teaches learned label matrix having albel-by-attention dimensions corresponding to ΒI∈RLI×da)), DI∈Rda×de (Vu, Page 3: "W is a matrix ∈ Rda×2u" (teaches learned projection matrix having attention- by hidden state dimensions, corresponding to DI∈Rda×de)); and
generating the intent label-specific matrix VI based on the intent-specific matrix EI and the intent label-specific attention weight matrix AI (Vu, Equation 6, (teaches generating a label specific matrix V from hidden state matrix H and attention matrix A, applying this to EI and AI yields VI.)).
Regarding claim 5, the combination of Zhu, Vu, and Lample discloses the method of claim 4.
Vu further discloses:
wherein the generating, by the label attention component, the slot label-specific matrices VS,k based on the slot-specific matrix ES, comprises:
computing a slot label-specific attention weight matrix AS,k based on a following formula:
AS,k=softmax(BS,k×tanhDS,k×ES) (Vu, Equations 4 and 5, Page 3, (teaches computing a label-specific attention weight matrix using the same structure, where H maps to ES, W maps to DS,k, U maps to BS,k, and A maps to AS,k.);
wherein the softmax is performed at a row level to make sure that a summation of weights in each row is equal to 1 (Vu, Page 3: "row level" and "summation of weights" (teaches row-level softmax normalization)), and ΒS,k∈RLS,k×da (Vu, Page 3: " U ∈ R|L|×da" (teaches learned label matrix having label-by-attention dimentions, corresponding to ΒS,k∈RLS,k×da)), DS,k∈Rda×de (Vu, Page 3: "W is a matrix ∈ Rda×2u" (teaches a learned projection matrix having attention by hidden-state dimension, corresponding to DS,k∈Rda×de)), at the kth hierarchy level, respectively (Vu, Abstract: "hierarchical relationships" (teaches labels organized by hierarchy level)); and
generating the slot label-specific matrices VS,k based on the slot-specific matrix ES and the slot label-specific attention weight matrix AS,k (Vu, Equation 6, (teaches generating a label-specific matrix V from hidden state matrix H and attention matrix A, applying this to ES and AS,k yields VS,k).
Vu does not explicitly disclose:
However, Zhu further discloses:
in which LI and LS,k are the intent label set and the set of slot label types (Zhu, P[0050]: "δ∈{I, S} (I represents the abbreviation of intent, S represents the abbreviation of slot), |δ| represents the number of predefined intent or slot labels" (teaches predefined intent and slot label sets))
Regarding claim 6, the combination of Zhu, Vu, and Lample discloses the method of claim 5.
Vu further discloses:
further comprising:
updating the slot label-specific matrix VS,k with a more coarse-grained label information (Vu, Abstract: "we additionally propose a hierarchical joint learning mechanism extending our label attention model to handle the issue, using the hierarchical relationships among the codes." (teaches use of hierarchical label information)) from a (k−1)th hierarchy level (Vu, Page 3, Figure 2 caption: "JointLAAT has two levels: The first level is to predict the normalized codes composing of the first three characters of raw ICD codes. The second level utilizes the prediction produced from the first level to predict the raw ICD codes." (teaches using a coarser first-level prediction to inform a finer second-level prediction, corresponding to information from a (k-1)th hierarchy level.)).
Regarding claim 13, claim 13 recites the spoken language (SLT) system corresponding to the methods described in claim 3 and is rejected under the same grounds as above.
Regarding claim 14, claim 14 recites the spoken language (SLT) system corresponding to the methods described in claim 4 and is rejected under the same grounds as above.
Regarding claim 15, claim 15 recites the spoken language (SLT) system corresponding to the methods described in claim 5 and is rejected under the same grounds as above.
Regarding claim 16, claim 16 recites the spoken language (SLT) system corresponding to the methods described in claim 6 and is rejected under the same grounds as above.
Allowable Subject Matter
Claims 7-10, and 17-20 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all the limitations of the base claims. While, Zhu does discuss a matrix used for slot filling and producing the predicted slot sequence (Zhu, P[0090] – P[0092]), a review of the prior art did not reveal the following limitations:
projecting each matrix Qt of the soft slot label matrix S, the intent label-specific matrix VI and the slot label-specific matrices VS,k into two spaces to obtain projected matrices Q←t and Q→t;
computing a bilinear attention between a previous matrix Qt-1 and a current matrix Qt to measure a correlation Ct between their corresponding label types;
generating the multiple intent representing matrix H←1 based on the projected matrices Q←t and Q→t and the correlation Ct.
While there were papers that described the general process of bilinear attention between intent and slot features, (such as Towards Joint Intent Detection and Slot Filling via Higher-order Attention by Dongsheng Chen et al.) they did not clearly disclose previous matrix Qt-1 and current matrix Qt in a previous/current ordered sequence as described in the claim limitations. The project and generating limitations above were also not found to be in any prior art found. As a result, claim 7 and its corresponding system claim 17 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all the limitations of the base claim. Claims 8-10, and 18-20 therefore, are also allowable if claims 7 and 17 were rewritten in independent form including all limitations of the base claim due to dependency on claims 7 and 17 respectively.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHASHIDHAR S MANOHARAN whose telephone number is (571)272-6772. The examiner can normally be reached M-F 8:00-4:00.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Andrew Flanders can be reached at 571-272-7516. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SHASHIDHAR SHANKAR MANOHARAN/Examiner, Art Unit 2655
/ANDREW C FLANDERS/Supervisory Patent Examiner, Art Unit 2655
Prosecution Insights